Developing a Solvent Extraction Process for the Separation of Cobalt and Iron from Nickel Sulfate Solutions

Home , Cobalt extraction

Abstract

DEVELOPING A SOLVENT EXTRACTION PROCESS FOR THE SEPARATION OF COBALT AND IRON FROM NICKEL SULFATE SOLUTIONS

Michiel Casparus Olivier

Thesis presented in partial fulfilment of the requirements for the Degree of

MASTER OF SCIENCE IN ENGINEERING (EXTRACTIVE METALLURGICAL ENGINEERING)

In the faculty of

Engineering at Stellenbosch University

Supervised by

Christie Dorfling

December 2011

Stellenbosch University http://scholar.sun.ac.za DeclarationAbstract

DECLARATION

By submitting this thesis electronically, I declare that the entirety of the work contained therein is my own, original work, that I am the sole author thereof (save to the extent explicitly otherwise stated), that reproduction and publication thereof by Stellenbosch University will not infringe any third party rights and that I have not previously in its entirety or in part submitted it for obtaining any qualification.

…………………………… ……………….

Signature Date

Stellenbosch University http://scholar.sun.ac.za Abstract

ABSTRACT

Crude NiSO4 solutions are often produced as a product of Sherrit based matte leach processes leading to iron and cobalt contaminated solutions of NiSO4. To upgrade the quality of these solutions for either, the production of NiSO4 crystals or cathode/precipitated nickel, the iron and cobalt must be removed.

Conventional processes use either pure or saponified Cyanex 272 in solvent extraction to extract iron and cobalt from pregnant nickel leach solutions. These processes require the addition of an alkali like NaOH to neutralise the protons being exchanged for the different metal species since extraction is a strong function of pH. Hence, while removing iron and cobalt from the solution, sodium is added instead. This limitation can be dealt with by pre-loading of Cyanex 272 with a portion of the purified nickel product prior to impurity extraction. During the extraction stage nickel is then exchanged for the impurities instead of hydrogen and no NaOH addition is necessary, resulting in a pure nickel product.

The extracted metals are recovered from the organic phase by stripping it with H2SO4 and the organic phase is recycled.

A solvent extraction process that purifies an 80 g/l nickel sulfate solution from 1 g/l cobalt and 3 g/l iron was developed. The key variables include pH, organic/aqueous (O/A) mixing ratio, nickel loading, H2SO4 concentration, Cyanex 272 concentration, various organic diluents and temperature. These variables were investigated via preliminary and three 24 factorial design batch experiments which reduced the variables to pH/NaOH addition and O/A ratio for the pre-loading section, nickel loading and O/A ratio for the extraction section and H2SO4 concentration and A/O ratio for the stripping section. The reduced variables were considered in further batch experiments from which the data was used to develop models and design a simulation sheet in Microsoft Excel of the extraction circuit. These final batch tests revealed the conditions that resulted in the purest aqueous nickel product after extraction, where nickel and cobalt can selectively be stripped from iron and what H2SO4 concentration and A/O ratio is necessary to finally strip iron and regenerate the organic phase. Ultimately four tests were conducted on a mixer-settler setup to test the validity of the simulation sheet.

The proposed process conditions for the solvent extraction circuit are to use a 20V% Cyanex 272 solution in Shellsol D70 as diluent that are pre-loaded with an 80 g/l purified nickel product solution at an O/A ratio of 25 and a pH of 7 generating a nickel loading of 3.2 g/l for the extraction section. This loading at an O/A ratio of 1.5 produces a high pure nickel aqueous product (0.05 g/l Co, 0.01 g/l Fe). From the generated organic phase co-extracted nickel and cobalt can selectively be stripped with 0.1 M

H2SO4 at an A/O ratio of 2.25 followed by an 1 M H2SO4 solution at an A/O ratio of 0.75 to strip the remaining iron. The temperature at all the stages should be controlled between 40-50°C to ensure phase separation.

Stellenbosch University http://scholar.sun.ac.za Opsomming

OPSOMMING

Ongesuiwerde NiSO4 oplossings word gedurig vervaardig as ŉ produk van die Sherrit gebaseerde mat logingsproses wat lei tot yster en kobalt gekontamineerder oplossings van NiSO4. Om die graad van hierdie oplossings op te gradeer vir die produksie van of NiSO4 kristalle of katode/gepresipiteerde nikkel, moet die yster en kobalt eers verwyder word.

Meeste prosesse gebruik of suiwer Cyanex 272 of die natrium sout van Cyanex 272 in ŉ vloeistof- vloeistof ekstraksie siklus om yster en kobalt vanuit versadigde nikkel oplossing te ekstraheer. Hierdie prosesse verg die gebruik van ŉ alkali soos NaOH om die protone te neutraliseer wat geruil word vir die verskillende metaal spesies, aangesien ekstraksie ŉ baie sterk funksie van pH is. Dit het tot gevolg dat natrium tot die oplossing bygevoeg word soos wat yster en kobalt verwyder word. Hierdie tekortkoming kan oorkom word deur Cyanex 272 eers te laai met nikkel deur ŉ porsie van die gesuiwerde nikkel produk te gebruik voor die onsuiwerhede geëkstraheer word. Gedurende die ekstraksie stadium word nikkel dan geruil vir die onsuiwerhede, in plaas van waterstof, en geen NaOH toevoeging is nodig nie. Die gevolg is ŉ suiwer nikkel oplossing. Die geëkstraheerde metaal word herwin vanaf die organies fase deur dit te stroop met ŉ H2SO4 oplossing en die organiese fase kan weer gehersirkuleer word.

ŉ Vloeistof-vloeistof ekstraksie proses wat ŉ 80 g/l nikkelsulfaat oplossing van 1 g/l kobalt en 3 g/l yster suiwer, is ontwikkel. Die belangrikste veranderlikes sluit in pH, organies/waterige (O/A) mengingsverhouding, nikkel lading, H2SO4 konsentrasie, Cyanex 272 konsentrasie, verskillende organiese oplosmiddels en temperatuur. Hierdie veranderlikes was ondersoek deur verskeie voorlopige en drie 24 faktoriaal ontwerp enkel-lading eksperimente wat gevolglik die veranderlikes verminder het na pH/NaOH toevoeging en O/A verhouding vir die voorbelading seksie, nikkel lading en O/A verhouding vir die ekstraksie seksie en H2SO4 konsentrasie en A/O verhouding vir die stroping seksie. Die verminderde veranderlikes was verder ondersoek in nog enkel-lading eksperimente, waarvan die data gebruik was om modelle op te stel en ŉ simulasie sigblad in Microsoft Excel te ontwerp van die ekstraksie siklus. Die finale enkel-lading toetse het die kondisies weergegee wat die mees suiwer waterige nikkel produk sal oplewer na ekstraksie, waar kobalt en nikkel selektief van yster gestroop kan word en watter H2SO4 konsentrasie en A/O verhouding nodig is om yster finaal te stroop en die organiese fase te regenereer. Laastens is vier toetse gedoen op ŉ menger-afskeidingstenk opstelling om die betroubaarheid van die simulasie sigblad te toets.

Die bedryfstoestande wat voorgestel word vir die vloeistof-vloeistof ekstraksie siklus is om ŉ 20V% Cyanex 272 oplossing in Shellsol D70 as oplosmiddel vooraf te laai met ŉ 80 g/l gesuiwerde nikkel produk oplossing by ŉ O/A verhouding van 25 en ŉ pH van 7 om ŉ nikkel lading van 3.2 g/l vir die ekstraksie seksie te genereer. Hierdie lading by ŉ O/A verhouding van 1.5 produseer ŉ suiwer nikkel waterige oplossing produk (0.05 g/l Co, 0.01 g/l Fe). Vanaf die gegenereerde organiese fase kan die geëkstraheerde nikkel en kobalt selektief gestroop word met 0.1 M H2SO4 by ŉ A/O verhouding van 2.25 gevolg deur ŉ 1 M H2SO4 oplossing by `n A/O verhouding van 0.75 om die oorblywende yster te stroop. Die temperatuur waarby al die stadiums beheer moet word is tussen 40-50°C om fase skeiding te verseker.

iii

Stellenbosch University http://scholar.sun.ac.za Acknowledgements

ACKNOWLEDGEMENTS

Galatians 2:20 (ESV): “I have been crucified with Christ. It is no longer I who live, but Christ who lives in me. And the life I now live in the flesh I live by faith in the Son of God, who loved me and gave himself for me”

First of all I would like to give all the Glory to Jesus Christ my Lord and Saviour. Without Him I can do nothing. He, through the Holy Spirit, is my primary source of favour, strength, endurance, knowledge and wisdom.

I would like to thank my parents, Niel and Marinda Olivier, for their financial and moral support. Without them I never would have had the opportunity to study Chemical/Metallurgical Engineering at the University of Stellenbosch which allowed me to do my Master of Science in Extractive Metallurgical Engineering.

To my fiancé Althea Viljoen, thank you for your love, respect, support and encouragement towards me. It strengthened me during the long hours I had to work on this project.

To my supervisor Christie Dorfling, thank you for your guidance, patience and input towards me as young engineer. Your excellent mentorship not only played a vital part towards the successful completion of this project, but also imparted valuable knowledge towards me as professional which I will be using throughout my career. For this I am sincerely grateful.

To Jacques Eksteen (Professor (Extraordinary) at the Department of Process Engineering (Stellenbosch) and consulting metallurgist at Western Platinum Limited, a subsidiary of Lonmin Plc), thank you for trusting me with this project and for your professional and technical guidance.

Thank you Nico Steenekamp (Metallurgist at Lonmin Plc) for your technical input.

And finally but not least, to Lonmin, without your financial support this project would not have been possible.

Stellenbosch University http://scholar.sun.ac.za Contents

CONTENTS

Abstract ...... ii Opsomming ...... iii Acknowledgements ...... iv Contents ...... v List of tables ...... ix List of Figures ...... xiii Nomenclature ...... xviii

Chapter 1

Introduction ...... 1

1.1 Background ...... 1 1.2 Problem statement ...... 2 1.3 Objective ...... 2

Chapter 2

Literature review ...... 3

2.1 Introduction ...... 3 2.2 General leach solution treatments ...... 3

2.2.1 Adsorption on activated charcoal ...... 3 2.2.2 Ion exchange ...... 3 2.2.3 Precipitation ...... 4 2.2.4 Solvent extraction ...... 4

2.3 Extracting Reagents ...... 5

2.3.1 Background ...... 5 2.3.2 DEHPA, PC-88A and Cyanex 272 ...... 7 2.3.3 Cyanex 272, 302 and 301 ...... 8

2.4 Diluents ...... 10 2.5 Phase separation & third phase formation ...... 11 2.6 Extraction and stripping chemistry ...... 11

2.6.1 General reaction ...... 11 2.6.2 Saponified Cyanex 272 ...... 13 2.6.3 Scrubbing ...... 14 2.6.4 Pre-loading ...... 14 2.6.5 Stripping ...... 15 2.6.6 Conclusions and technique selection ...... 16

Stellenbosch University http://scholar.sun.ac.za Contents

2.7 Operating variables ...... 17

2.7.1 pH control ...... 17

2.7.1.1 Base titrants ...... 18 2.7.1.2 Stripping agents/ Acid titrants ...... 19

2.7.2 Aqueous/Organic phase mixing ratio ...... 20

2.7.2.1 Background ...... 20 2.7.2.2 Number of stages for Ni, Co and Fe extraction ...... 22

2.7.3 Temperature ...... 24 2.7.4 Kinetics/Agitation time ...... 25

2.8 Pre-loading extraction ...... 25

Chapter 3

Batch tests ...... 28

3.1 Introduction ...... 28 3.2 Method ...... 29

3.2.1 Apparatus ...... 29 3.2.2 Solution preparation ...... 29 3.2.3 Experimental ...... 30

3.2.3.1 Pre-loading and Extraction ...... 31

3.2.3.1.1 Kinetics ...... 31 3.2.3.1.2 Comparing Cyanex 272 and Ni-Cyanex 272 ...... 32 3.2.3.1.3 Effect of various diluents ...... 32 3.2.3.1.4 Factorial designs ...... 33 3.2.3.1.5 O/A ratio, nickel loading and NaOH addition ...... 33

3.2.3.2 Stripping ...... 34

3.2.3.2.1 Kinetics ...... 34 3.2.3.2.2 Factorial design ...... 34 3.2.3.2.3 A/O ratio and acid concentration ...... 35

3.3 Results ...... 35

3.3.1 Pre-loading and Extraction ...... 35

3.3.1.1 Kinetics ...... 35 3.3.1.2 Comparing Cyanex 272 and Ni-Cyanex 272 ...... 37 3.3.1.3 Effect of various diluents ...... 40

Stellenbosch University http://scholar.sun.ac.za Contents

3.3.1.4 Factorial designs ...... 43

3.3.1.4.1 Pre-loading ...... 43 3.3.1.4.2 Extraction ...... 46

3.3.1.5 O/A ratio, nickel loading and NaOH addition ...... 49

3.3.2 Stripping ...... 54

3.3.2.1 Kinetics ...... 54 3.3.2.2 Factorial design ...... 54 3.3.2.3 A/O ratio and acid concentration ...... 56

Chapter 4

Modelling and Simulation ...... 59

4.1 Introduction ...... 59 4.2 Method ...... 59 4.3 Results ...... 62

4.3.1 Pre-loading ...... 63 4.3.2 Extraction ...... 64 4.3.3 Stripping ...... 66

4.4 Simulation ...... 67

4.4.1 Simulation sheet ...... 67 4.4.2 Model limitations ...... 68 4.4.3 Quantifying the number of stages ...... 72 4.4.4 Sensitivity analysis ...... 73

Chapter 5

Continuous tests ...... 76

5.1 Introduction ...... 76 5.2 Mixer-settler setup ...... 76 5.3 Mixer-settler configurations ...... 78 5.4 Experimental procedure ...... 80

5.4.1 Start up procedure ...... 80 5.4.2 Operating procedure ...... 82 5.4.3 Shut down procedure ...... 82

5.5 Results ...... 82

5.5.1 Extraction ...... 82

vii

Stellenbosch University http://scholar.sun.ac.za Contents

5.5.2 Stripping ...... 84

Chapter 6

Conclusions ...... 88

6.1 Literature review ...... 88 6.2 Batch tests ...... 88 6.3 Modelling and simulation ...... 90 6.4 Continuous tests ...... 91 6.5 Future work ...... 91

References ...... 93 Appendix A (Experimental and Modelling results) ...... 97

A1 Batch tests ...... 97 A2 Modelling and Simulation ...... 118 A3 Continuous tests ...... 124

Appendix B (Publications) ...... 129

viii

Stellenbosch University http://scholar.sun.ac.za List of tables

LIST OF TABLES

Chapter 1

Chapter 2

Table 2-1: Most popular organophosphorus extractants for the separation of cobalt, iron and nickel (adapted from (Flett, 2005)) ...... 6

Table 2-2: Effect of various diluents on extraction of cobalt (0.0025 g/l) with Cyanex 272 (0.005 M) at a pH of 8 (adapted from (Gandhi et al., 1993)) ...... 10

Table 2-3: Effect of various stripping agents for the stripping of cobalt from Cyanex 272 (adapted from (Gandhi et al., 1993)) ...... 19

Table 2-4: Effect of various stripping agents for the stripping of iron(III) from Cyanex 302 (adapted from (Ajgaonkar and Dhadke, 1997)) ...... 19

Table 2-5: Summary of the typical number of theoretical stages obtained when extracting cobalt and nickel with Cyanex 272 and stripping the organic with H2SO4 with various A/O ratios ...... 23

Table 2-6: Pre-loading Cyanex 272 (0.45 M) with 67.9 g/l of nickel at a pH of 5.97 and 6.27 (adapted from (O'Callaghan and Chamberlain, 2002)) ...... 26

Table 2-7: Extracting cobalt and other impurities from a nickel solution with the nickel salt of Cyanex 272 (adapted from (O'Callaghan and Chamberlain, 2002)) ...... 26

Chapter 3

Table 3-1: High and low values of the pre-loading and extraction 24 factorial designs (full design are in table A-9 in Appendix A) ...... 33

Table 3-2: High and low values of the stripping 24 factorial design (full design are in table A-19 in Appendix A) ...... 34

Table 3-3: Terms to express the repeatability values calculated with equation 21 (adapted from (Measey et al., 2003)) ...... 44

Table 3-4: Repeatability (r) values for the pre-loading factorial design ...... 44

Table 3-5: P-values for the pre-loading 24 factorial design with the high and low values shown in table 3- 1 ...... 44

Stellenbosch University http://scholar.sun.ac.za List of tables

Table 3-6: Repeatability (r) values for the extraction factorial design ...... 46

Table 3-7: P-values for the extraction 24 factorial design with the high and low values shown in table 3-1 ...... 47

Table 3-8: Repeatability (r) values for the stripping factorial design ...... 54

Table 3-9: P-values for the stripping 24 factorial design with the high and low values shown in table 3-2 ...... 55

Table 3-10: Phase separation and selectivity defined by the relative quantities stripped of each metal.. 56

Chapter 4

Table 4-1: R2 values of the models developed for equilibrium pH and nickel loading for the pre-loading section ...... 63

Table 4-2: R2 values of the models developed for equilibrium pH and cobalt and iron concentration in the aqueous phase at equilibrium ...... 65

Table 4-3: R2 values of the models developed for nickel, cobalt and iron concentration in the aqueous phase at equilibrium for the stripping section ...... 67

Table 4-4: Extrapolated model results for the pre-loading section ...... 69

Table 4-5: Extrapolated model results for the extraction section ...... 70

Table 4-6: Extrapolated model results for the stripping section ...... 71

Table 4-8: Iron concentration (g/l) in the organic phase at different stages as predicted by the simulation sheet to strip a 2 g/l iron loaded organic phase at an A/O ratio of 0.75 ...... 72

Table 4-9: Sensitivity analysis with respect to variations in the feed concentration to the extraction and stripping sections as predicted by the simulation sheet ...... 75

Chapter 5

Appendix A (Experimental and Modelling results)

Table A-1: Extraction kinetics of nickel, cobalt and iron with pure Cyanex 272 ...... 97

Stellenbosch University http://scholar.sun.ac.za List of tables

Table A-2: Pre-loading kinetics of Cyanex 272 with nickel ...... 97

Table A-3: Extraction kinetics of nickel, cobalt and iron with the nickel salt of Cyanex 272 ...... 98

Table A-4: Pre-loading of Cyanex 272 with different initial nickel aqueous concentrations...... 99

Table A-5: Sodium addition to nickel product solution as a function of cobalt and iron extraction with the nickel loadings of Cyanex 272 as generated in Table A-4 ...... 100

Table A-6: Sodium addition to nickel product solution as a function of cobalt and iron extraction with pure Cyanex 272 ...... 102

Table A-7: Effect of various diluents on nickel pre-loading ...... 103

Table A-8: Effect of various diluents on cobalt and iron extraction with Ni-Cyanex 272 ...... 104

Table A-9: High and low values of the pre-loading and extraction 24 factorial designs ...... 105

Table A-10: Results for the pre-loading factorial design as shown in Table A-9 ...... 106

Table A-11: Results for the extraction factorial design as shown in Table A-9 ...... 107

Table A-12: Effect of temperature on nickel loading of Cyanex 272 ...... 108

Table A-13: Effect of Cyanex 272 concentration on nickel loading of Cyanex 272 ...... 109

Table A-14: Effect of Cyanex 272 concentration on cobalt and iron extraction with Ni-Cyanex 272 ...... 110

Table A-15: Effect of temperature on the cobalt and iron extraction with Ni-Cyanex 272 ...... 111

Table A-16: Variation in O/A ratio and nickel loading for nickel, cobalt and iron extraction ...... 112

Table A-17: Variation in O/A ratio and NaOH addition for nickel loading of Cyanex 272 ...... 114

Table A-18: Data for the stripping kinetics of nickel, cobalt and iron ...... 114

Table A-19: High and low values of the stripping 24 factorial design ...... 115

Table A-20: Results for the stripping factorial design as shown in table 5-19 ...... 116

Table A-21: Variation in O/A ratio and aid concentration for nickel, cobalt and iron stripping ...... 117

Table A-22: Constants solved my minimizing the square of the error in Microsoft Excel for the models developed in the process simulations sheet ...... 118

Stellenbosch University http://scholar.sun.ac.za List of tables

Table A-23: Process conditions and results for the pre-loading (1st stage) and extraction (2nd stage) test when generating a lower nickel loading ...... 124

Table A-24: Process conditions and results for the pre-loading (1st stage) and extraction (2nd stage) test when generating a higher nickel loading...... 125

Table A-25: Process conditions and results when stripping the organic phase with the lower metal loading ...... 126

Table A-26: Process conditions and results when stripping the organic phase with the higher metal loading ...... 127

xii

Stellenbosch University http://scholar.sun.ac.za List of figures

LIST OF FIGURES

Chapter 1

Figure 1-1: Process flow sheet of a typical base metal refinery ...... 1

Chapter 2

Figure 2-1: Schematic illustration of a typical solvent extraction circuit utilizing cation exchangers ...... 5

Figure 2-2: Chemical structure of Cyanex 272, 301 and 302 (adapted from (Tait, 1993)) ...... 8

Figure 2-3: Schematic illustration of the difference between scrubbing and pre-loading ...... 15

Figure 2-4: Process flow sheet for the extraction of cobalt and iron from nickel solutions when using either pure or saponified Cyanex 272 ...... 16

Figure 2-5: Process flow sheet for the extraction of cobalt and iron from nickel solutions with Ni-Cyanex 272 ...... 17

Figure 2-6: Comparison of typical pH values at which Cyanex 272 extracts different metal species (adapted from (Habashi, 1999)) ...... 18

Figure 2-7: Example of a typical extraction isotherm constructed by varying the O/A ratio (adapted from (Tsakiridis and Agatzini-Leonardou, 2004)) ...... 20

Figure 2-8: Staged counter-current extraction system ...... 21

Figure 2-9: Example of a typical McCabe-Thiele diagram ...... 22

Figure 2-10: Effect of temperature on the extraction of cobalt (0.5 g/l) and nickel (0.5 g/l) with 0.025 M of Cyanex 272(adapted from (Sarangi et al., 1999b)) ...... 24

Figure 2-11: a: Extraction isotherms generated with 0.5 M Cyanex 272 and an initial nickel concentration of 80 g/l, b: extraction of cobalt (3 g/l) from an 83 g/l nickel solution at an A/O ratio of 1/5 with the generated nickel organic phase in Figure 2-11 (a) (adapted from (Nogueira et al., 2009)) ...... 27

Chapter 3

Figure 3-1: Schematic illustration of the order in which each set of batch experiments were conducted 31

xiii

Stellenbosch University http://scholar.sun.ac.za List of figures

Figure 3-2: Extraction kinetics of cobalt (1.05 g/l), iron (3.3 g/l) and nickel (88 g/l) with pure Cyanex 272 (see Table A-1 in Appendix A for the experimental data and conditions)...... 35

Figure 3-3: Pre-loading/Extraction kinetics of nickel (10.75 g/l) with pure Cyanex 272 (see Table A-2 in Appendix A for the experimental data and conditions) ...... 36

Figure 3-4: Extraction kinetics of cobalt (1.07 g/l), iron (3.3 g/l) and nickel (83.3 g/l) with Ni-Cyanex 272 (see Table A-3 in Appendix A for the experimental data and conditions)...... 36

Figure 3-5: Nickel loading with different initial nickel aqueous concentrations as a function of NaOH addition and pH (see Table A-4 in Appendix A for the experimental data and conditions) ...... 38

Figure 3-6: Extraction of cobalt (1 g/l) and iron (3 g/l) from the 80 g/l nickel solution as a function of nickel loading (see Table A-5 in Appendix A for the experimental data and conditions)...... 38

Figure 3-7: Extraction of cobalt (1 g/l), iron (3 g/l) and nickel (80 g/l) and sodium addition to the purified nickel product solution with pure Cyanex 272 (see Table A-6 in Appendix A for the experimental data and conditions) ...... 39

Figure 3-8: Extraction of cobalt (1 g/l), iron (3 g/l) and nickel (80 g/l) and sodium addition to the purified nickel product solution with the nickel salts of Cyanex 272 (see Table A-6 in Appendix A for the experimental data and conditions) ...... 40

Figure 3-9: Effect of using various diluents on pre-loading as a function of (a) pH with 10V% Cyanex 272 (b) NaOH addition with 10V% Cyanex 272 (c) pH with 20V% Cyanex 272 (d) NaOH addition with 20V% Cyanex 272 (see Table A-7 in Appendix A for the experimental data and conditions) ...... 41

Figure 3-10: Effect of using various diluents for cobalt (1.05 g/l), iron (2.99 g/l) and nickel (81.97 g/l) extraction with Ni-Cyanex 272 as a function of nickel loading with (a, c and e) 10V% Cyanex 272 and (b, d and f) 20V% Cyanex 272 (see Table A-8 in Appendix A for the experimental data and conditions) ...... 42

Figure 3-11: (a) Influence of temperature on nickel loading as a function of NaOH addition (b) Influence of extractant concentration on nickel loading as a function of NaOH addition (see tables A-12 and A-13 in Appendix A for the experimental data and conditions) ...... 45

Figure 3-12: Influence of extractant concentration on nickel loading as a function of pH (see tables A-12 and A-13 in Appendix A for the experimental data and conditions) ...... 46

Figure 3-13: Effect of extractant concentration on cobalt and iron extraction with Ni-Cyanex 272 (see Table A-14 in Appendix A for the experimental data and conditions) ...... 47

Figure 3-14: Effect of temperature on cobalt and iron extraction with Ni-Cyanex 272 (see Table A-15 in Appendix A for the experimental data and conditions) ...... 48

xiv

Stellenbosch University http://scholar.sun.ac.za List of figures

Figure 3-15: Nickel (80 g/l) extraction from and sodium addition to the product solution after extraction as a function of nickel loading and O/A (see Table A-16 in Appendix A for the experimental data and conditions) ...... 49

Figure 3-16: Variation in O/A ratio and nickel loading for pH and cobalt (1 g/l) and iron (3 g/l) extraction (see Table A-16 in Appendix A for the experimental data and conditions)...... 51

Figure 3-17: Variation in O/A ratio and pH with respect to nickel loading of Cyanex 272 (see Table A-17 in Appendix A for the experimental data and conditions) ...... 53

Figure 3-18: Stripping kinetics of an organic loaded with nickel (4.5 g/l), cobalt (1.04 g/l) and iron (2.96 g/l) (see Table A-18 in Appendix A for the experimental data and conditions) ...... 54

Figure 3-19: Nickel (2.69 g/l), cobalt (0.75 g/l) and iron (2.08 g/l) stripping as a function of H2SO4 concentration and A/O ratio (see Table A-21 in Appendix A for the experimental data and conditions) . 57

Chapter 4

Figure 4-1: Characteristic S shape of the sigmoid function ...... 60

Figure 4-2: Model results for nickel loading and equilibrium pH for the pre-loading section ...... 63

Figure 4-3: Model results for equilibrium pH and cobalt and iron concentration in the pregnant nickel solution at equilibrium for the extraction section ...... 65

Figure 4-4: Model results for nickel, cobalt and iron concentration in the aqueous phase at equilibrium for the stripping section ...... 66

Figure 4-5: Process simulation sheet in Microsoft Excel ...... 68

Figure 4-6: McCabe-Thiele diagrams for nickel loadings of 2.94 g/l and 3.95 g/l with an O/A feed ratio of 1.5 (a) and 2 (b) to extract 4 g/l and 5.5 g/l of iron ...... 73

Figure 4-7: McCabe-Thiele diagram for stripping 2 g/l and 5 g/l of nickel with 0.1 M H2SO4 at an A/O feed ratio of 2.25 ...... 74

Chapter 5

Figure 5-1: Schematic illustration of the organic and aqueous flow through the mixers and settlers ...... 77

Figure 5-2: Mixer-settler configuration for pre-loading and extraction when generating a lower nickel loading on the organic phase (process conditions and results are shown in Table A-23 in Appendix A) .. 78

Stellenbosch University http://scholar.sun.ac.za List of figures

Figure 5-3: Mixer-settler configuration for pre-loading and extraction when generating a higher nickel loading on the organic phase (process conditions and results are shown in Table A-24 in Appendix A) .. 78

Figure 5-4: Mixer-settler configuration when stripping an organic phase with a lower metal loading (process conditions and results are shown in Table A-25 in Appendix A) ...... 79

Figure 5-5: Mixer-settler configuration when stripping an organic phase with a higher metal loading (process conditions and results are shown in Table A-26 in Appendix A) ...... 79

Figure 5-6: Results for pre-loading (1st stage (a)) and extraction (2nd stage (b)) when generating a lower nickel loading on the organic phase (process conditions and results are shown in Table A-23 in Appendix A) ...... 82

Figure 5-7: Results for pre-loading (1st stage (a)) and extraction (2nd stage (b)) when generating a higher nickel loading on the organic phase (process conditions and results are shown in Table A-24 in Appendix A) ...... 83

Figure 5-8: Results for stripping the organic phase with the lower metal loading; (a) and (b) represent the 1st and 2nd stages respectively (process conditions and results are shown in Table A-25 in Appendix A) ...... 85

Figure 5-9: Results for stripping the organic phase with the higher metal loading; (a) and (b) represent the 1st and 2nd stages respectively (process conditions and results are shown in Table A-26 in Appendix A) ...... 85

Appendix A (Experimental and Modelling results)

Figure A-1: Three dimensional graphs of the models developed for pH and nickel loading for the pre- loading section ...... 118

Figure A-2: Three dimensional graphs of the models developed for pH and cobalt and iron concentration in the aqueous phase at equilibrium for the extraction section ...... 119

Figure A-3: Three dimensional graphs of the models developed for nickel, cobalt and iron concentration at equilibrium in the aqueous phase for the stripping section ...... 119

Figure A-4: Model results at various O/A ratios as a function of NaOH addition for pH and nickel loading at equilibrium for the pre-loading section ...... 120

Figure A-5: Model results at various nickel loadings as a function of O/A ratio for pH and cobalt and iron concentration in the aqueous phase at equilibrium for the extraction section ...... 122

Figure A-6: Model results at various A/O ratios as a function of H2SO4 concentration for nickel, cobalt and iron concentration in the aqueous phase at equilibrium for the stripping section ...... 123

xvi

Stellenbosch University http://scholar.sun.ac.za List of figures

Figure A-7: Photograph of the mixer-settler setup ...... 128

Figure A-8: Photograph of the agitators, mixers, settlers and aqueous overflow weirs ...... 128

xvii

Stellenbosch University http://scholar.sun.ac.za Nomenclature

NOMENCLATURE

Symbol Description Unit A/O Aqueous/organic mixing ratio (notation for stripping) ml/ml

b1-6 Modelling constants - β Separation factor - C Modelling constant -

Coaq Cobalt concentration in the aqueous phase at equilibrium g/l

Coo Cobalt concentration in the aqueous phase before mixing g/l

Coorg Cobalt concentration in the organic phase at equilibrium g/l D Distribution coefficient [g/l]/[g/l] E Percentage extracted % Loading pH, extraction pH, cobalt and iron extraction and nickel, f(y,x) -, % cobalt and iron stripping

Faq Aqueous flow rate ml/min fi Modelled value used to determine R2 g/l

FNaOH NaOH flow rate ml/min

Forg Organic flow rate ml/min

Feaq Iron concentration in the aqueous phase at equilibrium g/l

Feo Iron concentration in the aqueous phase before mixing g/l

Feorg Iron concentration in the organic phase at equilibrium g/l K Equilibrium constant (K) - [M] Metal concentration in the aqueous phase at equilibrium g/l

[M]o Metal concentration in the aqueous phase before mixing g/l

[M]org Metal concentration in the organic phase at equilibrium g/l 2 MSbetween Mean squares between groups (g/l) 2 MSwithin Mean squares within groups (g/l) n Number of repeated measurements -

NA Moles of metal A in aqueous phase after extraction mol

NAO Initial moles of metal A in the aqueous phase mol Relative moles of NaOH added with respect to the initial moles of NaOH/Nio mol/mol nickel in the aqueous phase

Niaq Nickel concentration in the aqueous phase at equilibrium g/l

Nio Nickel concentration in the aqueous phase before mixing g/l

Niorg Nickel concentration in the organic phase at equilibrium g/l Organic/aqueous mixing ratio (notation for pre-loading and O/A ml/ml extraction)

pH50 pH at which 50% of the metal species is extracted - r Repeatability - R2 Model performance parameter - S Percentage stripped %

xviii

Stellenbosch University http://scholar.sun.ac.za Nomenclature

2 SSerr Sum of squares of residuals (g/l) 2 SStot Total sum of squares (g/l)

Vaq Volume of the aqueous phase ml

Vorg Volume of the organic phase ml Represent NaOH/Nio (pre-loading), O/A ratio (extraction) and H SO mol/mol, ml/ml, x 2 4 concentration (stripping) in the respective models M Represent O/A ratio (pre-loading), nickel loading (extraction) and Y ml/ml, g/l, ml/ml A/O ratio (stripping) in the respective models y Average of the experimental values used to determine R2 -, g/l 2 yi Experimental value used to determine R -, g/l

xix

Stellenbosch University http://scholar.sun.ac.za Introduction

CHAPTER 1

1 INTRODUCTION

1.1 BACKGROUND

The mined ore being processed for PGM (platinum group metal) production typically consists of base metals such as nickel, copper, cobalt, and iron. Nickel, copper and cobalt are present in economically recoverable quantities while iron should be removed as an impurity. In order to recover the PGMs and base metals, the ore goes through various processing stages which usually include comminution, flotation, smelting and base metal refining.

Figure 1-1 gives a broad overview of a typical base metal refinery.

Nickel Fe and Co crystallizers Atmospheric leach separation 1st stage leach Thickener Milling Matte

Pressure leach 2nd stage leach Filters Fe and Co Caustic Nickel sulphate leach crystals

Formic leach Copper cathodes

Copper Electrowinning PGM concentrate Thickener

Figure 1-1: Process flow sheet of a typical base metal refinery

The matte from the smelter is ground in a closed circuit ball mill before being sent to the atmospheric leaching stage. During the first stage of atmospheric leaching nickel is dissolved from the nickel alloy and nickel sulphide phases in the matte using a sulphuric acid solution. The product stream from the first stage of leaching is sent to a thickener that separates the pregnant leach solution (rich in nickel) and the solid residue (rich in copper and PGMs). The overflow (pregnant solution) from the thickener is fed to a nickel crystallizer to produce nickel sulfate crystals as a hexa-hydrate in a rotary kiln (iron and cobalt removal prior to being fed to the crystallizer is discussed in sections 1.2 and 1.3). The residue from the

Stellenbosch University http://scholar.sun.ac.za Introduction

first stage thickener goes to the second stage of pressure leaching with the primary objective of leaching copper into solution, leaving the PGMs in the residue. Copper is recovered with electrowinning to produce copper cathodes. Filtering, caustic- and formic acid leaching are used to upgrade the PGM residue before it is sent to the precious metal refinery.

1.2 PROBLEM STATEMENT

As expected, some dissolution of copper, cobalt, iron and PGMs also occur during the atmospheric leaching stage. Consequently these metals are also present in the feed stream to the nickel crystallizer.

Precipitation processes are typically implemented to remove copper and PGMs from the leach solution prior to it being sent to the nickel crystallizer. The cobalt and iron present in the crystallizer feed stream do however reduce the quality of the crystals being produced. If cobalt and iron can be removed from the solution prior being sent to the nickel crystallizer, higher purity crystals can be produced allowing these to be sold at premium price. The recovered cobalt also has significant economic value.

1.3 OBJECTIVE

This project addresses the abovementioned deficiency by determining a feasible separation process and operating conditions to remove cobalt and iron from the pregnant nickel leach solution (with a composition of 1 g/l Co2+, 3 g/l Fe2+ and 80 g/l Ni2+) prior to being sent to the nickel crystallizer.

This was done by developing a solvent extraction circuit via the following approach:

 Preliminary and factorial design batch experiments were used to identify the key variables. These variables were considered during successive batch tests to determine the optimum operating conditions for the extraction circuit.  The generated equilibrium data was then modelled and a simulation sheet for the extraction circuit was developed.  Ultimately continuous tests were conducted to test the batch and simulated results.

Stellenbosch University http://scholar.sun.ac.za Literature review

CHAPTER 2

2 LITERATURE REVIEW

2.1 INTRODUCTION

The current literature survey considers typical leach solution treatments. Typical extractants used in solvent extraction for cobalt and iron separation from nickel sulfate solutions are discussed and a selection is made of the best competitor. These extractants are rarely used in pure form and diluents and other organics that are also involved in solvent extraction are discussed accompanied with several extraction techniques. The best technique is selected and discussed in more depth. Important operating variables are also discussed as they will be investigated to design the solvent extraction circuit.

2.2 GENERAL LEACH SOLUTION TREATMENTS

Typical processes that are used to treat leach solutions include adsorption on activated charcoal, ion exchange, precipitation and solvent extraction.

2.2.1 ADSORPTION ON ACTIVATED CHARCOAL

Adsorption on activated charcoal is generally used for the extraction of gold and silver from cyanide solutions which are filtered through columns filled with granular activated charcoal. The metal species that needs to be removed adsorbs onto the activated charcoal. For this method maximum efficiency is obtained for dilute solutions, the kinetics is slow, and it has a relatively low capacity and cost (Habashi, 1999).

The leach solution that has to be treated in the present project is highly concentrated with nickel making this an unattractive approach that could cause blockages and regular overloading of nickel onto the charcoal.

2.2.2 ION EXCHANGE

Generally ion exchange is used for concentration of uranium from leach solutions as well as separation of individual members of lanthanides. An ion exchange resin comprises a matrix that has been functionalised either with positive or negative ions. The metal species to be extracted from the leach solution is exchanged for the ion on the resin. For this method maximum efficiency is obtained for dilute solutions, it has slow kinetics, is relatively expensive and has a high operating capacity (Habashi, 1999).

Stellenbosch University http://scholar.sun.ac.za Literature review

Ion exchange is also not a popular choice for the same reasons as activated charcoal since the leach solution is highly concentrated with nickel, leading to overloading of nickel onto the resin.

2.2.3 PRECIPITATION

Typical precipitation processes include removal of iron from zinc sulfate solutions and precipitation of aluminium (Habashi, 1999). In precipitation solid particles are formed by changing operating conditions such as pH, addition of other reactants and the temperature of the solution. Addition of other reactants or increasing the pH, which will require the addition of an alkali like NaOH, defeats the purpose of purifying the pregnant nickel leach solution from cobalt and iron since new impurities are then added instead. Consequently this is not a good option.

2.2.4 SOLVENT EXTRACTION

Solvent extraction can be used for the extraction of U, Cu, Zn, Be, B, V, Cr, Fe, Co, Ni, Zr, Hf, Nb, Ta, Mo, W, As, PGMs, Au, Th, Pu and lanthanides from aqueous solutions (Habashi, 1999).

During solvent extraction the leach solution (aqueous phase) is contacted with an immiscible organic phase in either column or mixer-settler reactors. During agitation the metal ion from the aqueous phase reacts with the organic reagent (extractant) to form an organometallic complex. Once equilibrium is reached the two phases are allowed to separate and the metal species is extracted from the aqueous phase.

In practice extractants are rarely used in pure form. They are usually mixed with organic solvents known as diluents which should also be immiscible in water. Diluents are used to improve physical properties such as the viscosity of the organic phase. In contrast they may influence the extracting power of the extractant (Habashi, 1999).

In order to recover the extracted metal species from the organic phase as well as for recycling purposes of the reagent, the metal species in the organic phase is stripped with relatively strong acids such as

H2SO4 or HCl. During the extraction process a hydrogen ion on the organic reagent is replaced with the metal species in the aqueous phase, but during the stripping process the hydrogen ion from the stripping agent (H2SO4 or HCl) replaces the metal species on the organic reagent. Hence during the stripping process the organic reagent is regenerated to its acidic form and recycled to serve as extracting reagent while the metal species is recovered for further processing. A schematic illustration of a typical solvent extraction circuit is shown in Figure 2-1:

Stellenbosch University http://scholar.sun.ac.za Literature review

(RH)org

(RnM)org

Extraction Stripping

(M,N)aq (H2SO4/HCl)aq

(N)aq (M)aq

Figure 2-1: Schematic illustration of a solvent extraction circuit utilizing cation exchangers

Referring to Figure 2-1 the dashed and solid lines represent the organic and aqueous phases, respectively. In this scenario metal M is separated from metal N using the organic reagent RH. R represents a hydrocarbon anion and H is the proton replacing M during the extraction process.

Solvent extraction has a moderate cost relative to adsorption on activated charcoal and ion exchange and maximum efficiency can be obtained for concentrated solutions, the kinetics is rapid and it has a high operating capacity (Habashi, 1999).

Most frequently solvent extraction is the preferred separation process for cobalt, iron and nickel. This statement is justified with the significant amount of literature that is available in this field (Parhi et al., 2008; Biswas et al., 2007; Darvishi et al., 2005; Lindell et al., 2000; Devi et al., 1998; Ajgaonkar and Dhadke, 1997; Gandhi et al., 1993; Tait, 1993; Miralles et al., 1992). Solvent extraction is also capable of tolerating higher metal concentrations than most of its competitors and its application range over a much wider range of different metal species (Habashi, 1999). It is therefore selected as the separation process for the extraction of iron and cobalt from the pregnant nickel leach solution.

2.3 EXTRACTING REAGENTS

2.3.1 BACKGROUND

In previous years the separation of cobalt and nickel presented problems due to their similar chemical and physical properties. Neither nickel nor cobalt forms anionic sulfato complexes which mean that they cannot be extracted with anion exchangers. Thus for nickel/cobalt separation in sulfate mediums a cation exchanger is required (Flett, 2005). The physical and chemical properties of iron and nickel differ much more than for cobalt and nickel and as a result the removal of iron from nickel will not be as challenging as for cobalt. The primary focus is therefore to find a suitable extractant for cobalt and nickel. The selected extractant will then be verified for the final selection with respect to its extractability towards iron from nickel sulfate solutions.

Solvent extraction technology dates back to the 1950s where the first major commercial plant was commissioned for the recovery of uranium as by-product from gold mining in South Africa. Following was the first copper solvent extraction plants in the late 1960s in Arizona and Zambia while the first Ni-

Stellenbosch University http://scholar.sun.ac.za Literature review

Co solvent extraction plant was commissioned in 1974 (Sole et al., 2005). Currently there are a wide range of extractants being used for the extraction and purification of metal containing aqueous solutions. Organophosphorus extractants have proven to be of particular importance for the selective separation of cobalt and nickel (Flett, 2005). Table 2-1 summarizes the acid alkyl phosphorous extractants (cationic exchangers) that have been most widely investigated for the extraction of cobalt, iron and nickel:

Table 2-1: Most popular organophosphorus extractants for the separation of cobalt, iron and nickel (adapted from (Flett, 2005))

Type Name Commercial uses

Uranium extraction, rare earth

Alkyl phosphoric acids DEHPA/C16H35O4P extraction, cobalt/nickel separation, zinc extraction, etc.

Cobalt/nickel separation, rare Alkyl phosphonic acids PC-88A or Ionquest 801/ C H O P 16 35 4 earth separation

Cyanex 272, Ionquest 290, LIX 272/(C8H17)2P(O)OH Cobalt/nickel separation, zinc

Alkyl phosphinic acids Cyanex 302/(C8H17)2P(S)OH and iron extraction, rare earth separation

Cyanex 301/(C8H17)2P(S)SH

A great deal of research were conducted up to the 1970s for the separation of cobalt and nickel with DEHPA, but at that stage the industry still remained eager for a selective extractant for cobalt and nickel (Ritcey et al., 1975; Flett, 1972). It was during the early 1980s that Cyanex 272 and its thio analogues Cyanex 302 and Cyanex 301, was investigated in conjunction with DEHPA as extractant for nickel-cobalt separation (Flett, 2005). In section 2.3.2 phosphoric, phosphonic and phosphinic acids will be investigated by looking at DEHPA, PC-88A and Cyanex 272 respectively. The difference between Cyanex 272, 302 and 301 will be discussed in section 2.3.3.

In order to understand and quantify the performance of various extractants it is important to introduce the distribution coefficient, separation factor and percentage extraction (Habashi, 1999).

[1]

[2]

[3]

Stellenbosch University http://scholar.sun.ac.za Literature review

where: DA, DB = Distribution coefficients of metals A and B, respectively, at equilibrium β = Separation factor

EA = Percentage extracted of metal A

NAO = Initial moles of metal A in the aqueous phase

NA = Moles of metal A in aqueous phase after extraction

A high distribution coefficient (D) designates a high extractability of metal ions from the aqueous phase. The separation factor reflects how selectively metals are being extracted from the aqueous phase (a larger separation factor signifies better selectivity).

The pH of the aqueous solution is one of the most influential variables with respect to selective extraction. Finding the optimum pH that gives the highest separation factor is essential for obtaining selective extraction since different metal species are extracted at different pH values (Sarangi et al., 1999b; Devi et al., 1998). A good extracting reagent is classified as one that can extract two metal species at two completely different pH values, i.e. one that gives a high and low distribution coefficient for metals A and B, respectively, at a specific pH.

Another performance measurement tool is the pH50 value. This is the pH at which 50% of the metal species is extracted from solution. A large difference between the pH50 values of two metal species Ni -Co (pH50 for example) implies that they can be separated selectively.

2.3.2 DEHPA, PC-88A AND CYANEX 272

Devi et al. (1994) investigated DEHPA, PC-88A and Cyanex 272 as extracting reagents for cobalt without the presence of nickel from sulfate mediums. Benzene was used as diluent and the experiments were conducted at an organic/aqueous (O/A) mixing ratio of 1. Cobalt (0.59 g/l) was extracted with various extractant concentrations (0.005-0.08 M). Practically 100% extraction was achieved for a reagent concentration of 0.04 M and the extraction behaviour of all 3 reagents was very similar giving the same extraction for cobalt as a function of extractant concentration. The pH at which these three reagents extracted cobalt was however different, with complete removal obtained at pH values of approximately 6, 6.5 and 7 for DEHPA, PC-88A and Cyanex 272, respectively. It was concluded that all 3 extracting reagents are equally capable of completely removing cobalt from aqueous solutions, i.e. give a high distribution coefficient. These experiments however do not report any valuable information regarding the ability of these three reagents to selectively separate cobalt from nickel.

Rodrigues and Mansur (2009) used DEHPA and Cyanex 272 in kerosene as diluent to extract cobalt (2 g/l) and nickel (14 g/l) in the presence of 1 g/l cadmium from a sulfate medium with an O/A mixing ratio Ni-Co of 1. The pH50 values for DEHPA and Cyanex 272 was approximately 1 and 1.5 respectively. Darvishi et al. (2005) extracted cobalt and nickel (both 5 g/l) with 0.6 M DEHPA and Cyanex 272 also in kerosene Ni-Co with an O/A ratio of 1. pH50 values of 0.5 and 3 were reported for DEHPA and Cyanex 272 respectively. Devi et al. (1998) used a 0.6 M solution of Cyanex 272, PC-88A and DEHPA in kerosene to Ni-Co extract cobalt and nickel (both 0.6 g/l) from a sulfate medium and obtained pH50 values of approximately 3, 1.5 and 0.5 respectively. Separation factors ranging between 4000-7000, 70-1700 and

Stellenbosch University http://scholar.sun.ac.za Literature review

6-10 for Cyanex 272, PC-88A and DEHPA, respectively, have also been reported for aqueous solution consisting of 0.5-2 g/l cobalt and nickel (Sarangi et al., 1999b; Devi et al., 1998). Hence it is clear that Cyanex 272 is a better separation agent than either PC-88A or DEHPA for dilute cobalt and nickel solutions.

All three of these phosphorus based extractants have proven to be of particular importance in the past for the separation of cobalt and nickel but only Cyanex 272 is capable of treating solutions consisting of Ni:Co ratios higher than 100:1. The separation of minor quantities of cobalt from concentrated nickel solutions only became plausible since the development of Cyanex 272 and its thio analogues Cyanex 302 and Cyanex 301 (Flett, 2005).

It is well reported in the literature that the effectiveness of cobalt and nickel separation increases from phosphoric acids (DEHPA) to, phosphonic acids (PC-88A) to phosphinic acids (Cyanex 272) (Van de Voorde et al., 2006). Hence the final reagent selection will be made between the three phosphinic acids Cyanex 272, Cyanex 302 or Cyanex 301.

2.3.3 CYANEX 272, 302 AND 301

Cyanex 272 is the best known reagent produced by Cytec. Figure 2-2 shows the chemical structure of Cyanex 272 and its thio analogues Cyanex 302 and Cyanex 301.

R O

Cyanex 272 P

R OH

R S CH3

R = CH C CH CH CH2 Cyanex 301 P 3 2

CH3 CH3 R SH

R S

Cyanex 302 P

R OH

Figure 2-2: Chemical structure of Cyanex 272, 301 and 302 (adapted from (Tait, 1993))

Tait (1993) investigated the extraction behaviour of Cyanex 272, 302 and 301 on cobalt/nickel separation. Initial experiments were conducted where all three reagents were used to treat an aqueous solution consisting of 1.18 g/l cobalt and 1.18 g/l nickel. A reagent concentration of 0.25 M in toluene was used with an O/A mixing ratio of 1. The author found that the pH at which extraction occurred for Ni-Co cobalt and nickel differs much more for Cyanex 272 and Cyanex 302 than for Cyanex 301 with pH50

Stellenbosch University http://scholar.sun.ac.za Literature review

values of approximately 2.5 compared to 1.5 respectively. The pH at which cobalt and nickel are extracted with Cyanex 301 is also much lower than for Cyanex 272 or 302 (1-3 compared to 4-7) which means that a stronger acid will be required to strip cobalt and nickel from the organic phase. It was therefore concluded by the author that Cyanex 272 and Cyanex 302 are more suitable extracting reagents than Cyanex 301 and another experiment was conducted by extracting cobalt (0.42 g/l) from a more concentrated nickel (22.3 g/l) solution with a reagent concentration of 0.14 M in toluene at an O/A ratio of 1. Under these experimental conditions the separation factors reported by Tait (1993) were 53 and 126 at a pH of 6 for Cyanex 272 and Cyanex 302 respectively. From these experimental results the author concluded that Cyanex 302 is a more effective separation reagent than Cyanex 272 for cobalt and nickel.

The ability of an extracting reagent to selectively separate different metal species is not the only factor that should be taken into consideration when selecting a reagent. Possible degradation of the extractant during operation should also be considered. It is important that the selected extractant remains stable during the solvent extraction cycle. Cyanex 301 can be used to extract cobalt and nickel at a lower pH than either Cyanex 272 or 302. Hence it is a stronger extracting reagent than either Cyanex 272 or 302. Cyanex 301, however, is not very stable and decomposes into elemental sulphur and Cyanex 272. Even though Cyanex 302 shows better separation characteristics than Cyanex 272 it also decomposes into elemental sulphur and Cyanex 272 during the solvent extraction cycle. Cyanex 272 on the other hand has been found to be very stable (Flett, 2005; Mihaylov, 2003).

Cyanex 272 is a weaker acid than any of its competitors (DEHPA, PC-88A, Cyanex 302 and Cyanex 301) and a weaker acid can therefore be used for stripping purposes.

Cyanex 272 is currently being used to produce more than 50% of the cobalt in the Western world. Most companies regard the use of Cyanex 272 as confidential which means that not a lot of plant and operating data are available. It is therefore concluded that Cyanex 272 is the obvious choice for extraction of cobalt from nickel solutions based on the following reasons:

 It is more stable than its thio analogues Cyanex 302 and Cyanex 301.  It offers much higher separation factors than either DEHPA or PC-88A.  It is the only reagent that can effectively separate Co from Ni at Co:Ni ratios as low as 1:100.  It is also the weakest acid of all the organophosphorus acids which means that it is the easiest to strip.

The extraction characteristics of Cyanex 272 towards iron will be discussed next.

Miralles et al. (1992) investigated the extraction of iron (5.6 g/l) with Cyanex 272 from sulfate, chloride and nitrate mediums with 0.1 M Cyanex 272 in Isopar-H as diluent at an O/A ratio of 1. During these tests iron was completely removed from all three mediums at a pH lower than 2.2. This was however without the presence of cobalt and nickel in the solution. (Gandhi et al., 1993) studied the sequential separation of iron, cobalt and nickel with Cyanex 272 dissolved in chloroform as diluent. It was found

Stellenbosch University http://scholar.sun.ac.za Literature review

that iron was selectively extracted at a pH of 3, cobalt at a pH of 8 and nickel was left in solution with 0.005 M Cyanex 272. The proposed method was reported to be simple, rapid and selective.

Based on the selective extractability of Cyanex 272 for iron from cobalt and nickel solutions as well as its superior separation characteristics for cobalt and nickel, it is concluded as the best choice of extractant for the current situation.

2.4 DILUENTS

As discussed previously, extractants are rarely used in pure form and are mixed with an organic solvent referred to as the diluent. Various diluents exist, both aromatic and aliphatic, and they range from straightforward molecules to complex mixtures. Gandhi et al. (1993) investigated some of the straightforward molecules most often used on laboratory scale as diluents for Cyanex 272 by extracting cobalt. The results are displayed in Table 2-2:

Table 2-2: Effect of various diluents on extraction of cobalt (0.0025 g/l) with Cyanex 272 (0.005 M) at a pH of 8 (adapted from (Gandhi et al., 1993))

Diluents % Extraction n-hexane 60.8 Carbon tetrachloride 99.9 Cyclohexane 20.0 Benzene 99.9 Toluene 99.9 Xylene 99.9 Chloroform 99.9 Dichloromethane 98.3 1,2 dichloroethane 77.5 Nitrobenzene 56.5

Diluents may have a significant influence on the extracting behaviour of the extractant as shown in Table 2-2. This is caused by various interactions that may exist between diluents and extractants. Some of these interactions include dipole-dipole interaction, pi-electron interaction, hydrogen bonding and cavity formation. Some diluents also assist in the polymerization of the extractant. This will affect the cation exchange process, making the hydrogen atom on the extractant less free to react (Mohapatra et al., 2007).

The commercial use of the diluents in Table 2-2 is often restricted due to environmental issues or economic considerations. Kerosene is one of the cheapest diluents that is often used on industrial scale. Major oil companies have also developed a range of diluents for commercial application in hydrometallurgy. These diluents typically consist of a mixture of naphthenic, paraffinic and aromatic hydrocarbons. Some of these companies include Exxon which produces the Escaid range of diluents (Escaid 100, Escaid 110, and Escaid 350) and Shell and Chevron with their range of diluents (Shellsol D series) (Foust et al., 2008).

Stellenbosch University http://scholar.sun.ac.za Literature review

The interaction between diluents and extractants is poorly understood; hence a diluent is usually selected based on experimental data (Foust et al., 2008).

2.5 PHASE SEPARATION & THIRD PHASE FORMATION

A serious problem encountered in solvent extraction is the formation of a third phase. This occurs due to solubility issues in the organic phase. What typically occurs is that during the phase separation stage, a third phase forms between the aqueous and organic interface (the third phase has an intermediate density between the aqueous and organic phase). This phase is primarily an organometallic complex-rich phase. Above this phase forms an organic phase that is rich in the diluent. From literature the following conclusions have been drawn regarding phase separation problems and third phase formation (Foust et al., 2008):

 Suspended solids often result in phase separation problems.  Overloading of the ligand/extractant can result in third phase formation.  Third phase formation is temperature dependent. Increasing the temperature usually causes the third phase to disappear since the organic-aqueous solubility decreases as temperature increases.  Third phase formation is more common with aliphatic diluents.  Addition of a third phase modifier can help to overcome the formation of a third phase.

Typical third phase modifiers include isodecanol, 2-ethylhexanol, p-nonyl phenol and TBP (tri-n- butylphosphate). Most researchers use TBP as third phase modifier adding it between 2-5 volume percent to the organic phase (Rodrigues and Mansur, 2009; Reddy et al., 2006; Swain et al., 2006; Tsakiridis and Agatzini-Leonardou, 2004; Sarangi et al., 1999b; Devi et al., 1998).

Experimental evidence has shown that, like diluents, a third phase modifier is not chosen arbitrarily but a selection is made based on experimental observation (Foust et al., 2008). Most often an increase in temperature or extractant concentration causes the third phase to disappear and a third phase modifier is not necessary.

2.6 EXTRACTION AND STRIPPING CHEMISTRY

This section covers different extraction techniques associated with Cyanex 272 by specifically looking at the various chemical reactions. Restrictions linked with each technique will also be discussed. Finally a conclusion is drawn regarding which procedure will be the best choice for the purification of the pregnant nickel solution.

2.6.1 GENERAL REACTION

During solvent extraction a metal species is removed from the aqueous phase with a cation exchange reaction. If the interactions between diluents and extractants are ignored, this cation exchange reaction is presented with reaction 4 (Tsakiridis and Agatzini-Leonardou, 2004):

Stellenbosch University http://scholar.sun.ac.za Literature review

[4] where Mn+ is an n-valent metal cation, RH is a hydrocarbon (extractant) with proton H and subscripts “org” and “aq” represents the organic and aqueous phases respectively. Reaction 4 can be rewritten for the extraction of cobalt with Cyanex 272 as follows:

[5]

The release of hydrogen atoms into the aqueous phase during extraction will cause a drop in the pH of the aqueous solution. As discussed previously, the extraction of various components is a strong function of pH. Consequently the pH has to be controlled by adding OH- to the system during operation (typically + NaOH or NH4OH). This will prevent excess H formation that shifts the equilibrium of reaction 4 to the left. An improved representation for reaction 4 is shown in reaction 6 (a similar reaction can be written for NH4OH) (Nogueira et al., 2009):

[6]

With a better understanding of the extraction procedure associated with Cyanex 272, the actual chemistry in the system will be discussed in more depth by taking the interactions between the extractant and diluent into consideration. The following reaction is a more general representation of how Cyanex 272 extracts different metal species from aqueous solutions (Flett, 2005):

[7]

Reaction 7 can be rewritten where either NaOH or NH4OH are used as base titrants:

[8]

A significant amount of work has been done to determine the nature of the organometallic complex

MRn.mRH. Reaction 7 can be simplified to the following reaction for the extraction of cobalt and nickel with Cyanex 272 (Sahu et al., 2004; Tsakiridis and Agatzini-Leonardou, 2004):

[9]

Slope analysis is a technique usually used to determine the nature of the organometallic complex

MR2.m(RH)2. This is done via the definition of the equilibrium constant given by equation 10:

[10]

Stellenbosch University http://scholar.sun.ac.za Literature review

By taking the logarithm and after some rearrangement of equation 10 the following expression is obtained (Tsakiridis and Agatzini-Leonardou, 2004):

[11]

where DM is the distribution coefficient of metal M and [(RH)2] is the concentration of the extractant. A plot of (logDM–2pH) versus log[(RH)2] will yield a straight line of slope m and intercept logK. When m is known, the nature of the organometallic complex MR2.m(RH)2 is exposed. This has been reported as

CoA2.(HR)2 and NiR2.2(HR)2 for the cobalt and nickel respectively (Yu et al., 2001; Nogueira et al., 2009). With m known reaction 9 simplifies to the following two expressions for the extraction of cobalt and nickel:

[12]

[13]

The following simplification of reaction 7 is suggested by Van de Voorde et al., (2007) for ferric extraction with Cyanex 272 (reaction 14):

[14]

No research work has been discovered for the extraction of ferrous ions with Cyanex 272. It is however speculated that the reaction will be similar to reaction 9 due to the fact that ferrous ions also have a valence of 2 like cobalt and nickel.

It is therefore concluded that the chemistry involved with solvent extraction is not as simple as shown in reaction 4. This is caused by the interactions that exist between diluents and extractants as well as dimerization of extractants.

2.6.2 SAPONIFIED CYANEX 272

An alternative approach to using the acidic form of Cyanex 272 is to first prepare the sodium or ammonia salt of Cyanex 272 (saponified form of Cyanex 272) before extracting cobalt or iron. This will reduce the amount of sodium or ammonia hydroxide that has to be added to the extractor (extraction reactor) as shown in reactions 6 and 8 to control the pH. The two reactions involved with the extraction of divalent metal species (Co2+, Ni2+, Fe2+) with saponified Cyanex 272 are as follows (Kang et al., 2009; Reddy et al., 2006; Swain et al., 2006):

[15]

[16]

Stellenbosch University http://scholar.sun.ac.za Literature review

From reaction 16 it is seen that extraction still occurs via a cation exchange reaction, the only difference + + + is that the metal being extracted is exchanged for Na or NH4 rather than H . By using this technique the large drop in pH as a function of extraction is significantly reduced and hence the amount of alkali that has to be added during the extraction process is reduced. From reactions 15 and 16 it is also seen that the acidic form of Cyanex 272 exists as a dimer (RH2) and the saponified form as a monomer ((Na,

NH4)R)) (Kang et al., 2009; Reddy et al., 2006; Swain et al., 2006).

2.6.3 SCRUBBING

As expected, complete separation of two different metal species is not always achieved and the undesired metal is often co-extracted with the desired one to a certain extent. One way to remove the undesired metal from the organic phase is to mix it with an aqueous solution consisting of the desired metal.

Say for instance cobalt and nickel are extracted with Cyanex 272. In order to remove any co-extracted nickel from the organic phase, it is mixed with an aqueous solution of cobalt which scrubs nickel from the organic phase according to reaction 17 (Nogueira et al., 2009; Nogueira et al., 2003):

[17]

During scrubbing, only metal species with a higher chemical affinity for the extractant than the co- extracted metal can be used to scrub the undesired metal species from the organic phase. This is reflected in the pH at which Cyanex 272 extracts different metal species. The lower the pH at which Cyanex 272 extracts the metal species is, the higher the affinity for extraction will be with Cyanex 272. This means that metal species being extracted at lower pH values can be used to scrub/replace metal species that are extracted at higher pH values.

2.6.4 PRE-LOADING

Pre-loading is similar to scrubbing with respect to the chemistry as shown in reaction 17. The major difference between scrubbing and pre-loading is as follows:

 During scrubbing undesired and co-extracted nickel is replaced with cobalt in the organic phase by mixing the impure Ni-Co organic phase with a cobalt aqueous solution and in the ideal case will yield a pure Co-organic and a Ni-aqueous phase.  During pre-loading an organic phase is first loaded with nickel. The nickel loaded organic phase is then used to remove cobalt from a cobalt-nickel aqueous phase also according to reaction 17.

Figure 2-3 shows a schematic illustration of the difference between scrubbing and pre-loading.

Stellenbosch University http://scholar.sun.ac.za Literature review

NaOH

Co, Ni Cyanex Cyanex 272 272 Co-Cyanex 272 Scrubbing (a) after Extraction Scrubbing extraction Ni, Co Co Ni

Ni, Na

NaOH

Co-Cyanex 272 Pre-loading Cyanex 272 Ni-Cyanex 272 (b) before Pre-loading Extraction extraction Ni Ni, Co Ni

Figure 2-3: Schematic illustration of the difference between scrubbing and pre-loading

The technique shown in Figure 2-3 (a) will typically be used when a high purity cobalt product is of primary importance and the pre-loading technique in Figure 2-3 (b) will be used if a high purity nickel product is required. These two processes can be combined as well, but this is not of particular importance and will therefore not be discussed.

Pre-loading is based on the same principle as scrubbing regarding the fact that metal species being extracted at a higher pH value can be replaced in the organic phase with metal species being extracted at a lower pH value. Nickel, cobalt and iron are typically extracted in the pH range 7-8, 5-6 and 2-3, respectively, with Cyanex 272. Therefore, if a nickel salt of Cyanex 272 is agitated with an aqueous solution consisting of cobalt, iron and nickel, then iron and cobalt will replace nickel in the organic phase and increase the nickel concentration in the aqueous phase.

2.6.5 STRIPPING

Once the metal species is extracted into the organic phase, it has to be recovered via a process called stripping. Stripping is basically the inverse of extraction and involves mixing the metal loaded organic with an acid. It is the same as reaction 7, but only in reverse as shown in reaction 18:

[18]

During stripping Cyanex 272 is regenerated to its acidic form and a new aqueous phase is produced consisting of the extracted metal species. Since iron, cobalt and nickel are extracted at different pH values, a possibility exist that they can also selectively be stripped by using different acid concentrations (selective stripping). If iron and cobalt are extracted from a concentrated nickel solution for instance, then a weaker acid may be used to strip cobalt followed by a stronger acid to strip iron since iron is extracted at a lower pH than cobalt. Limited research regarding selective stripping is however available.

Stellenbosch University http://scholar.sun.ac.za Literature review

Flett (2005) was the only author that reported that cobalt can selectively be stripped from any co- extracted iron or zinc with 150 g/l H2SO4 (0.6 M) in Cyanex 272.

2.6.6 CONCLUSIONS AND TECHNIQUE SELECTION

To date most research has focused on either using pure Cyanex 272 (pure Cyanex 272 refers to Cyanex 272 that has not been pre-loaded with nickel for the remainder of the text) or the saponified form of Cyanex 272 to extract cobalt, iron and nickel from aqueous solutions. Saponified Cyanex 272 will typically be preferred when large quantities of metal have to be extracted since this will minimize

NH4OH/NaOH additions to the extractor unit making pH control easier (it will however require a separate unit for the preparation of saponified Cyanex 272). Figure 2-4 shows how such a process flow sheet will typically look like for the selective separation of nickel, cobalt and iron (potential scrubbing that may be required is not included):

NaOH

Saponification Cyanex 272 process

NaOH H2SO4 H2SO4 Cyanex 272 or Na-Cyanex 272 Co, Fe -Cyanex 272 Fe -Cyanex 272

Extraction of cobalt Stripping cobalt with Stripping iron with a and iron a relatively weak acid relatively strong acid Ni, Co, Fe

Ni, Na

Figure 2-4: Process flow sheet for the extraction of cobalt and iron from nickel solutions when using either pure or saponified Cyanex 272

It is observed that according to reactions 6 and 16 the extraction of 1 mole of metal M with valence n+ + + will release n moles of Na or NH4 into the aqueous solution. If the objective for example is to remove 1 2+ 2+ + + mole of Co and 1 mole of Fe from a concentrated nickel solution, then 4 moles of Na or NH4 will be added to the nickel solution. Thus while removing Co2+ and Fe+2 as impurities, double the amount of Na+ + or NH4 is added as a new impurity during the extraction process. The objective of the project is to develop a solvent extraction circuit that will produce a high purity nickel product, but this is not achieved via the flow sheet shown in Figure 2-4.

An alternative solution is to pre-load the organic phase with nickel according to reaction 13 and then use the nickel loaded organic to extract cobalt and iron from the concentrated nickel solution according + + to reaction 17 (this reaction will be similar for iron). According to reaction 17 no Na or NH4 is added to the purified nickel stream. The proposed flow sheet is shown in Figure 2-5:

Stellenbosch University http://scholar.sun.ac.za Literature review

Cyanex 272

NaOH H2SO4 H2SO4

Ni -Cyanex 272 Co, Fe -Cyanex 272 Fe -Cyanex 272 Extraction of Stripping cobalt Stripping iron Extraction of nickel/loading of with a relatively with a relatively cobalt and iron nickel Ni, Co, Fe weak acid strong acid

Co Ni Na

Figure 2-5: Process flow sheet for the extraction of cobalt and iron from nickel solutions with Ni-Cyanex 272

From Figure 2-5 a fraction of the purified nickel stream is recycled to be extracted by Cyanex 272. + + Consequently a nickel-loaded organic and a Na /NH4 salt is produced. The loaded organic is then used to remove cobalt and iron and stripping occurs in the same way as in Figure 2-4. An advantage of this approach is the fact that the purified nickel stream does not come into direct contact with the added sodium or ammonia. Any sodium or ammonia used is recovered as a sulfate salt and, in the ideal scenario these are pure enough so that no further treatment is required.

Finally it is concluded that pre-loading Cyanex 272 with nickel and then using the generated nickel salt of Cyanex 272 to extract cobalt and iron, is the preferred process route to purify the pregnant nickel leach solution.

2.7 OPERATING VARIABLES

With an understanding of the extraction and stripping chemistry involved in solvent extraction, several operating variables will be considered next.

2.7.1 PH CONTROL

As mentioned previously, pH is one of the most important operating variables. Figure 2-6 shows how Cyanex 272 can extract various metal species (not only nickel, cobalt and iron) at different pH values. The cures in Figure 2-6 are only illustrative and will vary depending on the operating conditions and relative metal ion concentrations.

Stellenbosch University http://scholar.sun.ac.za Literature review

100 2+ 2 3+ Zn2+ 3+ Cu Co2+ Ca2+ 90 Fe / Al Ni2+ 80

70 60 50 40

% Extracted % 30 20 10 0 0 1 2 3 4 5 6 7 8 pH

Figure 2-6: Comparison of typical pH values at which Cyanex 272 extracts different metal species (adapted from (Habashi, 1999))

Ni –Co Co-Fe Referring to Figure 2-6, Cyanex 272 has a pH50 value of 3 for nickel and cobalt and a pH50 value of 2 for cobalt and iron. This demonstrates the high selectivity that Cyanex 272 has for these three Co –Cu metals compared to copper and cobalt for example with a pH50 value of only 0.5. Nonetheless only iron, cobalt and nickel are of interest and will be considered.

Typical base and acid titrants that are used to control pH in solvent extraction will be considered next. Base and acid titrants are used for extraction and stripping respectively (base titrants increases the pH so that there is extraction and acid titrants/stripping agents decreases the pH so that there is stripping).

2.7.1.1 BASE TITRANTS

In solvent extraction it is important that the solution pH does not increase above the precipitation point of the metal species in solution to prevent the formation of any double salts (Konishi and Takano, 2002). When using the nickel pre-loading technique there is no direct alkali addition to the cobalt and iron contaminated pregnant nickel leach solution. A base is only added to the recycled nickel stream that is used to pre-load Cyanex 272 (refer to Figure 2-5). Hence double salt formation is only a potential problem during the nickel pre-loading stage and not during the extraction stage. Operating below a pH of 7.5 at this unit will prevent any nickel double salt (Ni(OH)2) formation (Nogueira et al., 2009; Park and Mohapatra, 2006).

NaOH and NH4OH are common base titrants used in solvent extraction. Nogueira et al. (2009) studied the effect of NaOH and NH4OH as base titrants for the extraction of cobalt and cadmium from nickel solutions with nickel pre-loaded Cyanex 272 and DEHPA (nickel salt of Cyanex 272 and DEHPA), respectively. It was reported that the use of ammonia led to the formation of nickel amine precipitates. The used of NaOH was subsequently preferred as the neutralizing agent since none of the precipitation issues were observed as with ammonia.

Stellenbosch University http://scholar.sun.ac.za Literature review

+ An alternative base to use instead of NaOH or NH4OH is NiCO3 which can be used to remove H according to reaction 19:

[19]

+ + If reaction 19 is used to control pH, it prevents the addition of any new cations such as Na or NH4 into the system. Nickel carbonate can then directly be used to control the pH and no nickel pre-loading will be required while only water addition/formation occurs to the purified nickel product stream and the

CO2 gas can be purged off.

After modelling the solubility behaviour of NiCO3 with the OLI analyzer software package, it was + observer that NiCO3 is only soluble in water at pH levels below 4 (i.e., there has to be significant H present in a system for NiCO3 to even dissolve). pH levels above 4 are required for complete cobalt removal with Cyanex 272. It is therefore concluded that NaOH as base titrant in a nickel pre-loading system is the best alternative.

2.7.1.2 STRIPPING AGENTS / ACID TITRANTS

Different stripping agents for cobalt and iron will be discussed next. Gandhi et al. (1993) considered various stripping agents for the removal of cobalt from Cyanex 272 solutions as shown in Table 2-3.

Table 2-3: Effect of various stripping agents for the stripping of cobalt from 0.005 M Cyanex 272 (adapted from (Gandhi et al., 1993))

% recovery Stripping agent (M) 0.01 0.05 0.1 0.5 1-5 HCl 57 71 91.3 99.9 99.9

HNO3 47.1 67.3 99.2 99.9 99.9

H2SO4 71.4 87.6 96.9 99.9 99.9

HClO4 73.1 86 98 99.9 99.9

CH3COOH 18.4 55.1 66.1 96 99.9

No stripping information for iron(II) from Cyanex 272 was discovered. Ajgaonkar and Dhadke (1997) however studied iron(III) stripping from a Cyanex 302 solutions. The results are displayed in Table 2-4:

Table 2-4: Effect of various stripping agents for the stripping of iron(III) from 0.005M Cyanex 302 (adapted from (Ajgaonkar and Dhadke, 1997))

% recovery Stripping agent (M) 0.5 1 1.5 2 3 HCl 91.8 99.7 99.7 99.7 99.7

H2SO4 77.7 85 99.8 99.9 99.7

HNO3 87 99.7 - 99.8 99.5

HClO4 99.8 99.8 - 99.8 -

Stellenbosch University http://scholar.sun.ac.za Literature review

From Tables 2-3 and 2-4 higher acid concentrations are required for stripping of iron than for stripping of cobalt. When considering H2SO4, it is observed that more than 96% recovery is achieved for concentrations of 0.1 M and 1.5 M for cobalt and iron respectively. A slightly lower concentration than 1.5 M will be expected for iron since Cyanex 272 is a weaker acid than Cyanex 302. No data with respect to iron stripping with acid concentrations less than 0.5 M were reported. Consequently it is unclear what percentage stripping might be expected at lower acid concentrations. As mentioned in section 2.6.5, Flett (2005) was the only author who stated, without any published data, that cobalt can selectively be stripped from any co-extracted iron or zinc with 150 g/l H2SO4 (0.6 M) from Cyanex 272. It was not reported whether this was for ferric or ferrous ions. No published data with respect to selectively stripping an organic phase consisting of cobalt and iron were discovered. This is therefore an area that still requires attention.

2.7.2 AQUEOUS/ORGANIC PHASE MIXING RATIO

2.7.2.1 BACKGROUND

The relative volumetric quantities at which the aqueous and organic phases are being mixed with is referred to as the aqueous/organic phase mixing ratio (A/O ratio).

Varying the A/O or O/A ratio at a fixed pH and initial aqueous concentration in batch experiments is typically used to construct extraction isotherms which are plots of the concentration of the metal species in the aqueous phase versus the concentration of the metal species in the organic phase at equilibrium. Figure 2-7 is a classic example of an extraction isotherm:

16 1/9 1/6 1/2 1/4 14 1/1

12 2/1

10 3/1 O/A 8 4/1

6 5/1 4 Metal content in organic phase (g/l) phase in organic content Metal 2 6/1

0 0 0.5 1 1.5 2 2.5 3 3.5 Metal content in aqueous phase (g/l)

Figure 2-7: Example of a typical extraction isotherm constructed by varying the O/A ratio (adapted from (Tsakiridis and Agatzini-Leonardou, 2004))

Stellenbosch University http://scholar.sun.ac.za Literature review

Say for instance the aqueous phase initially contains 5 g/l of a particular metal species. From Figure 2-7 at a relatively low O/A ratio (1/8) most of the metal is still in the aqueous phase at equilibrium (3.2 g/l). The corresponding high concentration on the organic axis (16.2 g/l) is due to the fact that 1.8g of metal 1 was transferred from 1l of an aqueous phase to /9l of an organic phase. At a high O/A ratio (6/1) most of the metal is transferred to the organic phase as shown on the aqueous axis. The corresponding concentration on the organic axis is not particularly high due to the fact that 4.75g of metal from a 1l aqueous phase has been transferred to 6l of an organic phase.

Therefore higher O/A ratios result in higher metal extraction and lower O/A ratios result in lower metal extraction. Frequently lower O/A ratios as well as operating pH values, where complete extraction in a single stage is not attained, are used. To reach complete extraction at these conditions a stepped wise process is implemented as shown in Figure 2-8:

yn+1 Forg y1 Forg y2 yn y3

n 2 1 x1 Xn-1 x2

xn Faq xo Faq

Figure 2-8: Staged counter-current extraction system

The organic and aqueous phases can also be fed in a co-current fashion, but the counter current manner shown in Figure 2-8 has proven to be more effective (Habashi, 1999).

If Faq, Forg, y and x in Figure 2-8 represent the flow rate of the aqueous phase, the flow rate of the organic phase, the concentration of the metal species in the organic phase and the concentration of the metal species in the aqueous phase, respectively, the following straight line is derived via a material balance and some mathematical manipulation:

[20]

where y1 = Metal content in the organic phase leaving the system (extract)

xo = Metal content in aqueous phase that is to be extracted

xn = Metal content in the aqueous phase leaving the system (raffinate)

yn+1 = Metal content in the organic phase entering the system n = Number of theoretical stages

Faq/Forg = Slope of the operating line (ratio at which the aqueous and organic phases are fed to the system in Figure 2-8)

The straight line in equation 20, also referred to as the operating line, is used in conjunction with the extraction isotherm in Figure 2-7 to determine the number of theoretical stages required to decrease

Stellenbosch University http://scholar.sun.ac.za Literature review

the metal content in aqueous phase from xo to xn and to increase the metal content in the organic phase from yn+1 to y1 as shown in Figure 2-9. y -

y1 1

(g/l) Operating line

slope-Faq/Forg

yn+1 3 Metal content in phase in organic content Metal

Xn Xo Metal content in aqueous phase-x (g/l)

Figure 2-9: Example of a typical McCabe-Thiele diagram

From Figure 2-9 it is clear that three theoretical stages are required to reach the desired extraction as have been stepped off by the dotted lines for a specific Faq/Forg or A/O feed ratio. The number of stages can be decreased by decreasing Faq/Forg ratio (increasing the volume at which the organic phase is fed relative to the aqueous phase). This will cause a drop in the slope of the operating line and decrease the number of steps shown by the dotted lines.

Figure 2-9 is referred to as the McCabe-Thiele diagram and is used when designing the number of stages in a solvent extraction system at a specific pH. The methodology for determining the number of stages in stripping systems is similar to that for the extraction systems by using the McCabe-Thiele diagram.

2.7.2.2 NUMBER OF STAGES FOR NI, CO AND FE EXTRACTION

With a better understanding of the O/A feed ratio in a solvent extraction system and its implications on the number of theoretical stages, typical examples will be discussed with respect to nickel and cobalt extraction. No data regarding the number of stages required to remove iron with Cyanex 272 were discovered. Table 2-5 summarizes the number of stages calculated in literature when using Cyanex 272 with different O/A ratios and initial aqueous concentrations for cobalt and nickel:

Stellenbosch University http://scholar.sun.ac.za Literature review

Table 2-5: Summary of the typical number of theoretical stages obtained when extracting cobalt and nickel with Cyanex 272 and stripping the organic with H 2SO4 with various A/O ratios

Number of stages for more Author Specifications A/O ratio than 98% extraction Extracting cobalt (0.59 g/l) in the presence of 0.59 g/l nickel from a 1 2 sulfate medium with 0.05 M of extractant Devi et al. (1998) Extracting nickel (0.59 g/l) from a sulfate medium with 0.05 M of 1 2 extractant Stripping of the loaded organic (Co and 0.5 2 Ni) with 0.02 M H2SO4 Extraction of cobalt (13.8 g/l) from a 2 g/l lithium and 15 mg/l nickel solution Kang et al. (2009) 0.5 2 from a sulfate medium with 0.4 M extractant Extraction of cobalt (1.78 g/l) from a Park and Mohapatra 16.78 g/l of nickel sulfate solution with 1.5 2 (2006) 0.2 M of the extractant Extraction of cobalt (0.63 g/l) in the presence of 3.8 g/l nickel and 5.75 g/l 2.5 2 Tsakiridis and Agatzini- magnesium with 0.54 M of the Leonardou (2004) extractant from a sulfate medium Extraction of nickel (3.8 g/l) with 0.54 M 2.4 2 of the extractant from a sulfate medium Extraction of cobalt (0.05 g/l) in the presence of 5.94 g/l of nickel with 0.03 2 2 M of the extractant from a sulfate Reddy et al. (2006) medium Stripping of cobalt (0.149 g/l) with 6 M 0.85 2 H2SO4 Extraction of cobalt (5.1 g/l) with 0.54 M Li et al. (2009) 1 2 of the extractant from a sulfate medium Extraction of nickel (3.6 g/l) with 0.27 M 2.4 2 of the extractant from a sulfate medium Agatzini-Leonardou et Stripping of nickel (8.6 g/l) with 2 M 2.4 3 al. (2009) H2SO4 Cobalt extraction (0.66 g/l) with 0.27 M 0.5 2 of the extractant from a sulfate medium Extraction of cobalt (22.08 g/l) in the presence of 5.43 g/l lithium with 1.3 M 1.5 2 Swain et al. (2008) Cyanex 272 from a sulfate medium Stripping of cobalt (44.719 g/l) with 3 M 0.4 3 H2SO4

Table 2-5 reflects the extensive amount of work that has already been done on the separation of cobalt and nickel with Cyanex 272. The small number of theoretical stages required for both extraction and

Stellenbosch University http://scholar.sun.ac.za Literature review

stripping (seldom larger than 2) proofs the superiority of Cyanex 272 with respect to cobalt and nickel refining. Most work reported on iron extraction is based on an O/A ratio of 1 while other variables such as extractant concentration and pH are varied. More than 90% of iron extraction was reported at an O/A ratio of 1(Biswas et al., 2007). It was therefore speculated that quantitative extraction of iron will also be possible in less than 3 stages.

2.7.3 TEMPERATURE

As discussed in section 2.5, temperature influences phase separation given that the solubility of the organic phase in the aqueous phase typically decreases as temperature increases. In this section extraction as a function of temperature will be examined.

Bhaskara Sarma and Reddy (2002) studied the extraction of nickel at macro-level concentrations (29.345 g/l) with Cyanex 272 within the temperature range 30 to 45°C. It was reported that temperature had no influence on nickel extraction within the given temperature range.

Darvishi et al. (2005) found that the extraction of cobalt (5 g/l) and nickel (5 g/l) from aqueous solutions is endothermic when using mixtures of DEHPA and Cyanex 272 (1:1) as extractant within the temperature range 25-60°C. It was reported that the separation factor of cobalt and nickel increases from 9.93 to 51.8 when the temperature increases from 25°C to 60°C. Increasing the temperature by 35°C decreased the pH at which cobalt is extracted to a larger extent than that of nickel (roughly 0.4 and 0.8 for nickel and cobalt respectively). This explains the increased separation factor as temperature increases.

Sarangi et al. (1999b) extracted cobalt (0.5 g/l) and nickel (0.5 g/l) within the temperature range as shown in Figure 2-10:

25 14

Distribution coefficient-Co 12 20 Distribution coefficient-Ni 10 Co/Ni (separation factor) 15 8

10 6 4

5 factor Separation

Distribution coefficient Distribution 2

0 0 10 15 20 25 30 35 40 45 50 T (oC)

Figure 2-30: Effect of temperature on the extraction of cobalt (0.5 g/l) and nickel (0.5 g/l) with 0.025 M of Cyanex 272(adapted from (Sarangi et al., 1999b))

From Figure 2-10 nickel extraction slightly increases as temperature increases. Cobalt extraction increases as temperature increases from 10 to 30°C and then decreases again within the temperature

Stellenbosch University http://scholar.sun.ac.za Literature review

range 30-50°C. It was reported that the decrease in cobalt extraction is caused by the fact that the organometallic complex becomes unstable at the elevated temperatures. The highest separation factor for the selective separation of cobalt and nickel (both 0.5 g/l) was obtained at 30°C. No temperature data with respect to iron extraction were discovered.

Based on this discussion, temperature only has a significant influence when extracting minor quantities of nickel and not when extracting major quantities of nickel. Increasing the temperature also increases the separation factor for cobalt and nickel (Darvishi et al., 2005). The stability issue of the cobalt extracted complex at higher temperatures mentioned by Sarangi et al. (1999b) was not reported by Darvishi et al. (2005) who extracted cobalt at temperatures up to 60°C. The stability of the cobalt extracted complex at temperatures above 30°C therefore remains unclear even though Sarangi et al., (1999b) and Darvishi et al. (2005) extracted metal species with different concentrations of 0.5 and 5 g/l respectively.

2.7.4 KINETICS/AGITATION TIME

The kinetics involved with cobalt, iron and nickel extraction will be focussed on in this section.

Hubicki and Hubicka (1996) and Lindell et al. (2000) investigated the extraction characteristics of cobalt (0.4 g/l and 3.5 g/l respectively) and nickel (58.7 g/l and 30 g/l respectively) with Cyanex 272 (0.27 M and 0.54 M respectively) as a function of contact time in a mechanical shaker and residence time in a co- current glass column reactor, respectively.

Hubicki and Hubicka (1996) found that equilibrium was reached after 90 seconds of agitation in a mechanical shaker. Lindell et al. (2000) stated that a residence time of 20 seconds in a glass column reactor was sufficient to reach equilibrium. A possible reason for the faster kinetics achieved in the latter case is that better mass transfer was obtained in the glass column reactor than in the mechanical shaker. In addition there are several authors that have reported that 5 minutes of agitation is sufficient to reach equilibrium for cobalt and nickel extraction and stripping (Parhi et al., 2008; Devi et al., 1998; Gandhi et al., 1993).

Slower kinetics for iron(III) than either cobalt(II) or nickel(II) extraction with Cyanex 272 was reported with contact times as high as 30 minutes being used (Biswas and Singha, 2006). The pregnant leach solution however consists of ferrous ions (Fe2+) and not ferric ions (Fe3+). No kinetic data are available concerning ferrous extraction with Cyanex 272 and is consequently poorly understood.

2.8 PRE-LOADING EXTRACTION

The selected nickel pre-loading technique to extract cobalt and iron from the pregnant nickel leach solution will be considered next. Three papers that used the nickel salts of the extractant to remove impurities from a concentrated nickel stream will be discussed.

Stellenbosch University http://scholar.sun.ac.za Literature review

The first paper is a patent from O'Callaghan and Chamberlain (2002). The method involved contacting Cyanex 272 with a portion of a purified nickel solution and then using the pre-equilibrated cationic extractant to purify the impure nickel solution wherein the impure metal species are exchanged for nickel from Cyanex 272 (refer to Figure 2-5). An experiment was conducted by O'Callaghan and Chamberlain (2002) where 0.45 M of Cyanex 272 diluted in kerosene was contacted with 67.9 g/l of a nickel solution containing less than 50 mg/l of cobalt at 50°C and an O/A ratio of 1 for 5 minutes. The pH was adjusted to 5.97 and 6.26 for test 1 and 2, respectively, generating two different nickel loaded organic phases as shown in Table 2-6.

Table 2-6: Pre-loading Cyanex 272 (0.45 M) with 67.9 g/l of nickel at a pH of 5.97 and 6.27 (adapted from (O'Callaghan and Chamberlain, 2002))

pH Aqueous Organic Ni content before extraction (g/l) 4.97 67.9 0 Ni content after extraction (test 1) (g/l) 5.97 60.4 7.5 Ni content after extraction (test 2) (g/l) 6.26 54.2 13.7

From Table 2-6 the two organic phases that were generated that consisted of 7.5 and 13.7 g/l of nickel for test 1 and 2 respectively. The pre-equilibrated organic phase was then contacted with an impure nickel solution consisting of 71.98 g/l Ni, 6.5 g/l Co, 11 mg/l Cu, 5 mg/l Zn, and 11 mg/l Mn at 50°C and an O/A ratio of 1 for 5 minutes. The results from tests 1 and 2 are shown in Table 2-7.

Table 2-7: Extracting cobalt and other impurities from a nickel solution with the nickel salt of Cyanex 272 (adapted from (O'Callaghan and Chamberlain, 2002))

Ni Co Cu Zn Mn Before extraction (g/l) 71.98 6.5 0.011 0.005 0.011

After extraction (g/l) 78.1 0.085 0 0 0 Test 1 % extracted -8.62 98.69 100 100 100 After extraction (g/l) 79.72 0.04 0 0 0 Test 2 % extracted -10.88 99.38 100 100 100

Referring to Table 2-7 most of the impurities are removed from the nickel solution with only trace quantities of cobalt remaining after extraction for both nickel loadings. The negative extraction of nickel designates nickel addition to the nickel solution. It was reported that the current invention is not limited to the experimental setup discussed above, but can be used to remove any impurity metal species that are extracted at a lower pH than nickel with Cyanex 272.

The second paper focussed on separating cobalt and cadmium from nickel sulfate solutions with the nickel salts of Cyanex 272 and DEHPA respectively (Nogueira et al., 2009). Only cobalt separation with Cyanex 272 will be discussed since this is of primary concern. Typical leach liquors obtained from Ni-Cd batteries consists of 80 g/l Ni, 30 g/l Cd and 3 g/l Co. After cadmium removal with DEHPA the liquor consisted of 80 g/l Ni, 0.1 g/l Cd and 3 g/l Co which was the composition of the liquor that was treated with the nickel salt of Cyanex 272 by Nogueira et al. (2009).

Stellenbosch University http://scholar.sun.ac.za Literature review

An experiment was conducted where Cyanex 272 (0.5 M) was equilibrated with 80 g/l of a nickel solution at various O/A ratios and two different pH values (the extraction isotherms generated are displayed in Figure 2-11 (a)). The generated nickel organic phase was then used to remove cobalt from the leach liquor at a constant O/A ratio of 5 (Figure 2-11 (b))

13 11 12 10 11 9 10 8

9 7 (g/l) 8 pH = 6 6 pH = 6 7 pH = 6.3 5 pH = 6.3 Nickel in organic organic in (g/l) Nickel 6 4 20 30 40 50 60 70 80 6 7 8 9 10 11 12 13 Cobalt in resulting organic in resulting Cobalt Nickel in aqueous (g/l) Nickel in organic (g/l) (a) (b)

Figure 2-41: a: Extraction isotherms generated with 0.5 M Cyanex 272 and an initial nickel concentration of 80 g/l, b: extraction of cobalt (3 g/l) from an 83 g/l nickel solution at an A/O ratio of 1/5 with the generated nickel organic phase in Figure 2-11 (a) (adapted from (Nogueira et al., 2009))

Referring to Figure 2-11 (b), increasing the nickel content in the organic phase causes an increase in cobalt extraction. This is also observed from Tables 2-6 and 2-7 where a higher nickel loading removed more cobalt from the aqueous phase. It is therefore concluded that nickel pre-loading can be used for the removal of cobalt from nickel solutions.

The third paper focussed on purifying a 100 g/l nickel solution from Ca and Mg with the nickel salt of a 20 V% DEHPA solution (Clark et al., 1993). The purified raffinate contained less than 100 mg/l Mg and less than 5 mg/l Ca. The nickel salt of DEHPA can therefore also be used to purify the nickel solution. Cyanex 272 is however still selected as extractant since it is more selective than DEHPA with respect to cobalt- nickel separation.

No nickel pre-loading work with respect to removing iron from nickel solutions was discovered. It is however speculated that the pre-loading technique will also work for iron since iron is extracted at a lower pH that either cobalt or nickel.

The primary reason for rather using pre-loaded (Figure 2-5) than pure Cyanex 272 (Figure 2-4) is to minimize sodium additions to the nickel product stream. None of the authors reported any data with respect to sodium additions to the nickel product stream and only characterized the extraction behaviour of cobalt and other impurities with the nickel salt of the extractant. All the remaining papers that were reviewed only stated whether NaOH or NH4OH were used to control the aqueous pH and what the concentration of these base titrants was. None of them reported any data with respect to the sodium or ammonia addition to the nickel product solution after extraction. A comparison of the relative sodium addition to the aqueous product solution after extraction when using Ni-Cyanex 272 or pure Cyanex 272 could therefore not be made.

Stellenbosch University http://scholar.sun.ac.za Batch tests

CHAPTER 3

3 BATCH TESTS

3.1 INTRODUCTION

The objective of the batch experiments was to determine the operating variables for a solvent extraction circuit that will separate 1 g/l and 3 g/l of cobalt and iron, respectively, from an 80 g/l nickel sulfate solution by using the nickel-preloading technique. These variables included the following:

 pH/NaOH addition, O/A ratio and temperature for the pre-loading section  nickel loading, O/A ratio and temperature for the extraction section

 H2SO4 concentration, A/O ratio and temperature for the stripping section

As expected, the same Cyanex 272 concentration and type of diluent will be used at each section as selected from the experimental data. Various batch experiments were conducted in a specific order since each set of experiments revealed essential information to continue with successive tests.

Initially the extraction kinetics was considered. After the minimum time required to reach equilibrium had been verified from the kinetic tests, the sodium addition to the purified nickel solution was investigated when using pure Cyanex 272 and when using the nickel salt of Cyanex 272. Once it was certain that using the nickel salt of Cyanex 272 is the preferred process route, various diluents were evaluated to identify any interactions that may exist between the extractant and diluents. After selecting a diluent, two 24 factorial designs for the pre-loading and extraction sections were conducted to discern the key and trivial variables as well as to identify any interactions there may exist between the design variables. Finally the O/A ratio was varied for the pre-loading and extraction sections with various NaOH additions and nickel loadings, respectively. From these tests the operating conditions for both sections were determined.

Next, stripping tests were carried out to see if selective stripping of cobalt, iron and nickel from the organic phase is possible as well as to see what operating conditions will be required to completely regenerate the extractant to its acidic form. Initially the stripping kinetics were considered to determine the minimum agitation periods for the rest of the experiments where after another 24 factorial design was performed to distinguish the vital and insignificant variables as well as to identify any important parameter interactions. Lastly the A/O ratio and acid concentration were varied simultaneously which revealed which operating conditions are needed for selective stripping of nickel, cobalt and iron.

Stellenbosch University http://scholar.sun.ac.za Batch tests

3.2 METHOD

Following is a discussion of how the apparatus and solutions were used and prepared as well as the objectives of each set of experiments.

3.2.1 APPARATUS

Most published work with respect to cobalt, iron and nickel extraction with Cyanex 272 is conducted in a batch fashion. The organic/aqueous dispersion is either mixed via a wrist shaking action (Gandhi et al., 1993) or with a mechanical agitator (Darvishi et al., 2005) where after separation funnels are used to allow for phase separation (Parhi et al., 2008). The metal content in the aqueous phase before and after each experimental run is most frequently determined via atomic absorption spectrophotometry (AAS). Inductively coupled plasma mass spectrometry (ICP-MS) was also used in some cases (Lothongkum et al., 2009), but the majority of papers (Nogueira et al., 2009; Reddy et al., 2006; Rodrigues and Mansur, 2009; Sahu et al., 2004; Tsakiridis and Agatzini-Leonardou, 2004) have reported that AAS is the preferred analytical technique used for analysis of the aqueous phase. ICPMS will often be preferred for very dilute solutions (parts per billion) since AAS is only capable of detecting metal concentrations as low as parts per million. The metal content in the organic phase was either determined via a mass balance or by stripping the organic phase with a very strong acid and then using AAS or ICP-MS to determine the metal content in the stripping agent afterwards by these authors.

In the current work three magnetic stirrer hot plates were used with either 100 ml or 500 ml beakers (depending on the dispersion volume) to serve as mixing reactors. The aqueous and organic solutions were preheated to the desired operating temperature by using a water heating bath. Pipettes were then used to transfer the organic and aqueous phases into the beakers. The hotplates were further used to control the temperature during agitation. An adjustable pipette was used to add NaOH in order to adjust the pH of the solution during agitation where necessary. After agitation was complete the pH was measured and the dispersion was transferred into 500 ml separation funnels allowing sufficient time for the two phases to separate. After phase separation was complete the aqueous phases were collected from the separation funnels and diluted to the concentration range required for analysis by AAS, and the samples were analyzed (0 – 100 mg/l). The metal content in the organic phase was determined via a mass balance from the AAS measurements.

3.2.2 SOLUTION PREPARATION

CoSO4.7H2O, NiSO4.6H2O and FeSO4.7H2O (CP grade) were supplied by KIMIX South Africa and were dissolved in distilled water to prepare synthetic representations of the pregnant leach solution and the aqueous solution used for nickel pre-loading. The metal contents in the synthetic solutions were verified via AAS.

Cyanex 272 was supplied by Cytec Inc. and was used without further purification. The concentration of Cyanex 272 required in the organic phase is dependent on the concentration of the metal species in the

Stellenbosch University http://scholar.sun.ac.za Batch tests

aqueous phase. As expected, larger quantities of metal species in the aqueous phase will demand a higher Cyanex 272 concentration in the organic phase. The active component/ligand of Cyanex 272 is bis(2,4,4-trimethylpentyl)phosphinic acid and constitutes 85% of Cyanex 272 (Hubicki and Hubicka, 1996). Hence a 10V% solution of Cyanex 272 in a particular diluent represents a 0.27 M solution of bis(2,4,4-trimethylpentyl)phosphinic acid and a 20V% Cyanex 272 solution represents a 0.54 M solution of bis(2,4,4-trimethylpentyl)phosphinic acid. It has been reported that Cyanex 272 should not be operated at conditions were the ligand is loaded to higher than 60% of its total capacity according to the chemical reactions discussed previously (reactions 12-14). It has been observed that exceeding 60% of the loading capacity of Cyanex 272 causes an increase in the viscosity of the organic phase (O'Callaghan and Chamberlain, 2002; Devi et al., 1994). An increase in viscosity will also cause phase separation issues which will complicate operating conditions.

The objective was to remove 1 g/l of Co and 3 g/l of Fe from an 80 g/l Ni solution. Estimating the concentration of the organic phase for an O/A ratio of 1 in a single stage will typically be as follows:

 According to reaction 17, 1 mole of a divalent ion is swapped for 1 mole of another divalent ion during the extraction stage. So if there is 1 g/l (0.017 M) and 3 g/l (0.054 M) of cobalt(II) and iron(II) in the impure nickel stream, respectively, then approximately 4.17 g/l (0.054 M + 0.017 M = 0.071 M) of nickel needs to be totally extracted during the pre-loading stage.  If reaction 4 is used as a first approximation then a 2x0.071 M = 0.142 M Cyanex 272 solution is required if the ligand is to be 100% loaded. This increases to 0.24 M if viscosity issues due to exceeding the 60% loading capacity of the ligand are taken into consideration.

An approximation of the minimum Cyanex 272 concentration that will prevent ligand overloading at an O/A ratio of 1 is therefore 10V% (0.27 M). A higher concentration will typically be required since the pre- loading reaction is also more likely to occur via reaction 13 rather than reaction 4.

H2SO4 was also supplied by KIMIX South Africa and was used as stripping agent.

3.2.3 EXPERIMENTAL

Figure 3-1 summarizes the order in which the batch tests were conducted and is followed by a more in depth discussion of each set of pre-loading, extraction and stripping tests.

Stellenbosch University http://scholar.sun.ac.za Batch tests

Pre-loading and Extraction

Compare pure Cyanex Consider various Kinetic tests 272 and Ni-Cyanex 272 24 factorial designs Variation in O/A ratio diluents mechanisms

Identify any major Discern between key and Determine minimum time Verify that the sodium Determine the optimum influences that diluents trivial variables and identify required to reach addition to the purified operating conditions for have on the extracting any interactions between equilibrium for pre-loading product solution decreases the pre-loading and power of Cyanex 272 for design variables for the pre- and extraction experiments when using Ni-Cyanex 272 extraction sections pre-loading and extraction loading and extraction sections

Stripping

Kinetic tests 24 factorial design Variation in O/A ratio

Discern between key and Determine minimum time trivial variables and identify Determine the optimum required to reach any interactions between operating conditions for equilibrium for the design variables for the the stripping section stripping section stripping section

Figure 3-1: Schematic illustration of the order in which each set of batch experiments were conducted

3.2.3.1 PRE-LOADING AND E XTRA CTION

3.2.3.1.1 Kinetics

Initially the extraction kinetics was investigated to determine the minimum agitation time for all succeeding extraction experiments. A 20V% (or 0.54 M) Cyanex 272 solution was used ensuring that the ligand is not overloaded resulting in viscosity and phase separation issues. An O/A ratio of 1 (20 ml/20 ml) was used throughout all the experimental runs at a temperature of 50°C.

Firstly the extraction kinetics of cobalt (1.05 g/l) and iron (3.3 g/l) from an 88 g/l nickel solution with pure Cyanex 272 (no pre-loading with nickel before extraction) in kerosene was investigated by allowing agitation to continue for 20, 40, 60, 100, 150 and 300 seconds. For each run 1 ml of a 3 M NaOH solution was added (approximately 1.95 mole of OH- for each mole of cobalt and iron that is initially in the aqueous phase).

Next the nickel loading kinetics of Cyanex 272 in kerosene was investigated. A 10.75 g/l nickel solution was used with a 3 M NaOH addition of 1 and 2 millilitres (there is approximately 0.8 and 1.6 moles of OH-, respectively, for each mole of nickel that is initially in the aqueous phase) at agitation time intervals of 20, 40, 60, 80, 100, 150, 200, 250 and 300 seconds.

Finally the extraction kinetics with Ni-Cyanex 272 in kerosene and Shellsol D70 was investigated. After verifying the minimum agitation time to load Cyanex 272 with nickel, four organic batches were prepared consisting of 2.47 g/l, 4.09 g/l, 4.66 g/l and 8.48 g/l of nickel. These four batches were then used, respectively, to extract cobalt (1.07 g/l) and iron (3.3 g/l) from an 83.3 g/l nickel sulfate solution at agitation time intervals of 20, 40, 60, 100, 150 and 300 seconds.

Stellenbosch University http://scholar.sun.ac.za Batch tests

3.2.3.1.2 Comparing Cyanex 272 and Ni-Cyanex 272

After verifying the extraction kinetics, two sets of experiments were conducted where cobalt (1 g/l) and iron (3 g/l) were extracted from a concentrated nickel (80 g/l) sulfate solution with pure Cyanex 272 as well as with the nickel salt of Cyanex 272 at various pH values/NaOH additions to the extraction circuit. The objective of these experiments was to determine to what extent using the nickel salt of Cyanex 272 decreases sodium additions to the purified nickel stream compared to using pure Cyanex 272. All the experiments were carried out at an A/O ratio of 1(20 ml/20 ml) and a temperature of 50°C.

For the case where pure Cyanex 272 was used (NaOH is directly added to the pregnant leach solution during extraction), a 20V% batch of Cyanex 272 in kerosene was prepared. The 3 M NaOH additions were 0.15, 0.35, 0.5, 0.65, 0.75, 0.85, 1, 1.3, 1.75, 2, 2.25 and 2.75 millilitres which are equivalent to 0.32, 0.74, 1.06, 1.37, 1.59, 1.80, 2.11, 2.75, 3.70, 4.23, 4.76 and 5.81 moles of OH-, respectively, for each mole of cobalt and iron that are initially in the aqueous phase. This was sufficient to obtain a range of experimental data points for extraction as a function of pH for cobalt and iron since a good first approximation would be to add 2 moles of OH- for each mole of a divalent ion that needs to be extracted as was seen previously.

For the scenario where the nickel salt of Cyanex 272 was first prepared prior to extraction (NaOH is only added in the unit where Cyanex 272 is converted into Ni-Cyanex 272 and not during the extraction stage), a 20V% batch of Cyanex 272 in kerosene was also prepared and were pre-loaded by using initial nickel aqueous concentrations of 6.15 g/l, 10.5 g/l and 21.2 g/l with 3 M NaOH additions of 0.25, 0.5, 0.75, 1, 1.3, 1.6, 2 and 2.25 millilitres each. The reason for using various initial nickel concentrations during the pre-loading stage was to see if higher initial nickel concentrations would result in higher loadings than lower initial nickel concentrations as a function of pH/NaOH addition. It will also be revealed whether using different initial nickel concentrations has an effect on cobalt and iron extraction or if only the final nickel loading on the organic phase is the determining factor of the extent of extraction.

3.2.3.1.3 Effect of various diluents

Next the effect of using various diluents was considered to identify any considerable influences they may have on the extracting power of the extractant. These results revealed whether a cautious decision should be made when selecting a diluent, or if economical considerations are the leading factor to consider.

The diluents that were used during the batch experiments included kerosene, heptane, toluene, xylene and Shellsol D70. Shellsol D70 is a 0% aromatic commercial diluent that is normally used with Cyanex 272 for the extraction of cobalt and nickel and was supplied by Shell Chemicals South Africa. The rest of the diluents were supplied by KIMIX South Africa and are CP grade.

10V% and 20V% Cyanex 272 solutions were prepared with all the diluents and was pre-loaded with a 10.3 g/l nickel solution with 3 M NaOH additions of 0.25, 0.5, 1 and 2 millilitres at an A/O ratio of 1 (20

Stellenbosch University http://scholar.sun.ac.za Batch tests

ml/20 ml) and 50°C. The nickel salts of Cyanex 272 were then afterwards used to extract cobalt (1.05 g/l) and iron (2.99 g/l) from an 81.97 g/l nickel sulfate solution.

3.2.3.1.4 Factorial designs

After selecting a diluent a 24 factorial design was conducted for both the pre-loading and extraction sections. During the pre-loading design the influences of A/O ratio, extractant concentration, NaOH addition and temperature on nickel loading (initial nickel aqueous concentration was 9.76 g/l), sodium addition to the aqueous phase after extraction and pH were investigated. During the extraction design the influences of A/O ratio, extractant concentration, nickel loading and temperature on extraction and pH were investigated. The aim of these two designs was to eliminate variables that do not have a significant effect for subsequent experimental work as well as to identify any interactions between these variables. Three replicates were carried out of each run to check the repeatability of the experimental setup. The high and low values for the two designs are shown in Table 3-1.

Table 3-1: High and low values of the pre-loading and extraction 24 factorial designs (full design are in Table A-9 in Appendix A)

Pre-loading O/A ratio Extractant concentration NaOH addition Temperature ml organic ml aqueous (V%) 3 M (ml) (°C) high 40 20 20 2 60 low 10 20 10 1 30 Extraction O/A ratio Extractant concentration Ni loading Temperature ml organic ml aqueous (V%) (g/l) (°C) high 40 20 20 4.76 60 low 10 20 10 2.83 30

Preliminary nickel pre-loading and extraction experiments were also conducted at 25°C with similar NaOH additions as the experiments conducted at 50°C (a 10V% and 20V% Cyanex 272 were used at both temperatures). These runs however presented various phase separation problems caused by solubility issues at these low temperatures as discussed during the literature survey. Nonetheless some data points could still be generated and were used in conjunction with the data obtained at 50°C to validate some of the claims made from the 24 factorial design with respect to the influence of temperature. A lower temperature of 30°C in the design was chosen to try and prevent solubility problems and a maximum of 60°C was chosen since the dispersion becomes too volatile at higher temperatures. The rest of the high and low values were chosen based the results obtained from the previous experiments.

3.2.3.1.5 O/A ratio, nickel loading and NaOH addition

Ultimately the O/A ratio and nickel loading were varied simultaneously to determine the number of theoretical stages required for the extraction section. As will be seen during the results and discussion section a 20V% Cyanex 272 solution were chosen at a temperature of 50°C for the rest of the extraction

Stellenbosch University http://scholar.sun.ac.za Batch tests

experiments. The O/A ratio was varied between 0.25, 0.5, 0.75, 1, 1.5, 2, 2.5 and 3 with nickel loadings of 0.56, 1.97, 2.94, 3.95, 4.69 and 5.73 g/l at each O/A ratio.

After selecting the O/A ratio and nickel loading for the extraction section, the O/A ratio and pH/NaOH addition for the pre-loading section were investigated. It was important to first determine the O/A ratio and nickel loading requirements for the extraction section before optimizing operating conditions for the pre-loading section. To prevent additional water addition to dilute the nickel pre-loading aqueous phase (or a portion of the recycled purified nickel aqueous phase), a very high O/A ratio will be required since an 80 g/l Ni stream will then be used instead of a 6-21.2 g/l Ni solution as was the case in the previous experiments. Based on mass balance calculations the O/A ratio for the pre-loading section ranged between 13.33 and 26.67 for nickel loadings of 6 g/l and 3 g/l respectively. Finally O/A ratios of 26.67, 20, 16 and 13.33 (80 ml/3 ml, 80 ml/4 ml, 80 ml/5 ml and 80 ml/6 ml respectively) and 3 M NaOH additions of 2.5, 3, 4, 5, 5.5 and 6 millilitres were selected as design variables to determine the number of theoretical stages and operating conditions for the pre-loading section. A 20V% Cyanex 272 solution was also used for these tests at 50°C as was concluded from previous experiments.

3.2.3.2 STRIPPING

3.2.3.2.1 Kinetics

The stripping kinetics was investigated next to determine minimum agitation periods to reach equilibrium. These results served as an indication for the rest of the stripping experiments regarding their agitation periods.

A 0.1 M and 0.5 M H2SO4 solution at agitation periods of 0.5, 1, 5, 7, 10 and 15 minutes were considered. An organic (20V% Cyanex 272 solution) composition of 4.5 g/l Ni, 1 g/l Co and 3 g/l Fe was used. The A/O ratio and temperature chosen were 1 and 50°C respectively.

3.2.3.2.2 Factorial design

Next a 24 factorial design was conducted for the stripping section as well. The influence of A/O ratio, acid concentration, temperature and the metal content of the organic phase on percentage stripped were investigated. The objective of the factorial design was to indentify the most important variables that would be taken into consideration in the next set of experiments. Each run were repeated three times to check for repeatability. The high and low values of the design are displayed in Table 3-2.

Table 3-2: High and low values of the stripping 2 4 factorial design (full design are in Table A-19 in Appendix A)

A/O ratio Acid concentration Metal loading Temperature ml organic ml aqueous (M) (g/l Ni, Co, Fe) (°C) high 20 10 0.5 4.67, 1.03, 3.14 60 low 20 40 0.1 2.26, 0.52, 1.58 30

Stellenbosch University http://scholar.sun.ac.za Batch tests

3.2.3.2.3 A/O ratio and acid concentration

Finally the A/O ratio and acid concentration was varied simultaneously to see if it would be possible to selectively strip nickel, cobalt and iron from the organic phase as well what the minimum acid concentration should be used to regenerate Cyanex 272 for recycling purposes. An A/O ratio of 0.5,

0.75, 1, 1.5, 2.25 and 3 with H2SO4 concentrations of 0.025, 0.05, 0.075, 0.1, 0.25, 0.5, 0.75, 1, 2 and 4 M were used during these tests.

3.3 RESULTS

3.3.1 PRE-LOADING AND EXTRACTION

The results for the pre-loading and extraction tests will be considered in the specific order as discussed in section 3.2.

3.3.1.1 KINETICS

The kinetic results for cobalt and iron extraction with pure Cyanex 272 are displayed in the Figure 3-2.

100 7

90 6 80 70 5 60 4

50 Ni pH 40 Co 3

Extraction (%) Extraction 30 Fe 2 20 pH 1 10 0 0 0 50 100 150 200 250 300 Agitation time (s)

Figure 3-2: Extraction kinetics of cobalt (1.05 g/l), iron (3.3 g/l) and nickel (88 g/l) with pure Cyanex 272 (see Table A-1 in Appendix A for the experimental data and conditions)

From Figure 3-2 equilibrium is reached within 100 seconds which compares well with the time required to reach equilibrium determined by Hubicki and Hubicka, (1996) (see section 2.7.4). Hubicki and Hubicka, (1996) had an aqueous solution consisting of 0.4 g/l cobalt and 58.7 g/l nickel; hence purification of a nickel sulfate solution was also of primary concern and confirms the kinetics in Figure 3- 2. It is observed that nickel extraction decreases over time. Referring to the pH, it is observed that as cobalt and iron is extracted, hydrogen is released into the aqueous phase causing the pH to drop. Assuming that some nickel was extracted within the first 20 seconds, the drop in pH can explain the drop in nickel extraction with time.

The nickel pre-loading kinetics of Cyanex 272 is displayed in Figure 3-3.

Stellenbosch University http://scholar.sun.ac.za Batch tests

65 Ni (NaOH/Nio = 0.82) 55 Ni (NaOH/Nio = 1.64)

Extraction (%) Extraction 45

35 0 50 100 150 200 250 300 Agitation time (s)

Figure 3-3: Pre-loading/Extraction kinetics of nickel (10.75 g/l) with pure Cyanex 272 (see Table A-2 in Appendix A for the experimental data and conditions)

As shown in Figure 3-3 the kinetics for pre-loading Cyanex 272 with nickel is faster than extraction of cobalt and iron with Cyanex 272 with equilibrium being reached in less than 20 seconds.

Figure 3-4 shows the kinetic results for nickel, cobalt and iron extraction with Ni-Cyanex 272.

10 6.0 8

5.5 6 4 5.0 2

4.5 0 pH -2 4.0 -4

3.5 (%) extraction Ni -6 -8 3.0 -10 0 50 100 150 200 250 300 0 50 100 150 200 250 300

Agitation time (s) Agitation time (s) (a) (b)

100 100

80 80

60 60

40 40

Fe extraction (%) extraction Fe 20 Co extraction (%)extraction Co

0 0 0 50 100 150 200 250 300 0 50 100 150 200 250 300

Agitation time (s) Agitation time (s) (c) (d)

Ni loading = 2.47g/l (shellsol D70) Ni loading = 4.66g/l (shellsol D70) Ni loading = 8.48g/l (kerosene) Ni loading = 4.09g/l (kerosene)

Figure 3-4: Extraction kinetics of cobalt (1.07 g/l), iron (3.3 g/l) and nickel (83.3 g/l) with Ni-Cyanex 272 (see Table A-3 in Appendix A for the experimental data and conditions)

Stellenbosch University http://scholar.sun.ac.za Batch tests

The kinetics with respect to cobalt and iron extraction with Ni-Cyanex are slower than when using pure Cyanex 272 with equilibrium being reached after 150 seconds compared to 100 seconds. The kinetic data regarding nickel extraction in Figure 3-4 (b) are scattered up to 150 seconds where after it becomes more stable. A possible explanation is that in the beginning the nickel cations in the organic phase are exchanged for cobalt and iron while the initial high pH also causes some nickel to be removed from the aqueous solution by the organic. The extracted nickel can then also be exchanged for cobalt and iron in the aqueous phase. The extent to which these reactions occur will depend on the nickel loading and pH as cobalt and iron removal proceeds.

The time to reach equilibrium decreases as the nickel loading on the organic phase increases for cobalt and iron extraction. This can be explained by considering typical rate laws that are functions of the reagent concentration/nickel loading. Hence lower and higher nickel loadings will decrease and increase the rate of cobalt and iron removal respectively. For lower nickel loadings, cobalt extraction initially increases and then starts decreasing again (Figure 3-4 (c)). A potential explanation can be given by considering the extraction chemistry and by looking at the behaviour of pH in Figure 3-4 (a). Initially the nickel in the organic phase is exchanged for cobalt and iron according to reaction 17. Once the nickel is exhausted from the organic phase the pH is still high enough for iron extraction to occur and the protons on the ligand are exchanged for iron according to reaction 7. This causes the pH to drop and cobalt is also exchanged from the organic phase into the aqueous phase for iron until iron becomes exhausted. For the case where a high nickel loading (8.48 g/l) was used there is sufficient nickel in the organic phase to be exchanged for both cobalt and iron. A drop in pH is however still observed but not to such an extent that cobalt is released back into the aqueous phase. This may also explain the faster kinetics observed for higher nickel loadings since extraction then mainly occurs via exchanging of nickel for cobalt and iron with the additional mechanism of cobalt being released back into the aqueous phase due to the drop in pH occurring to a lesser extent.

It was concluded that a total agitation period of 5 minutes will be allowed for successive extraction and pre-loading experiments. This corresponds well to the times used by several authors (Rodrigues and Mansur, 2009; Park and Mohapatra, 2006; Bhaskara Sarma and Reddy, 2002; Sarangi et al., 1999a; Park, 2007; Sarangi et al., 1999a).

3.3.1.2 COMPARING CYANEX 272 AND NI-CYANEX 272

Before comparing the sodium additions to the purified nickel stream after extraction, the effect of using different initial nickel concentrations during the pre-loading stage will be considered. The nickel loading of the organic phase after pre-loading with different nickel concentration aqueous solutions with varied NaOH additions and generated pH values are shown in Figure 3-5.

Stellenbosch University http://scholar.sun.ac.za Batch tests

12 12 10

8 8 6 6 organic (g/l) organic -

organic (g/l) organic 4

- 4 Ni Ni 2 2

0 0 0.0 0.5 1.0 1.5 2.0 2.5 5.2 5.7 6.2 6.7 7.2 7.7 8.2 pH ml of 3M NaOH for O/A = 20ml/20ml (b) (a)

pre200 -loading with 6.15g/l Ni pre-loading with 10.5g/l Ni pre-loading with 21.2g/l Ni i - r g N o a 0.0 ml of 3M NaOH0.5 for A/O = 20ml/20ml1.0 … 1.5 2.0 2.5

Figure 3-5: Nickel loading with different initial nickel aqueous concentrations as a function of NaOH addition and pH (see Table A-4 in Appendix A for the experimental data and conditions)

From Figure 3-5 (a) it is clear that the loading of nickel onto the organic phase remains similar for all three initial nickel concentrations at the same NaOH additions. Loading with the 6.15 g/l nickel solution is lower than the 10.5 g/l and 21.2 g/l solutions at 3 M NaOH additions higher than approximate 1.25 ml. This is due to the fact that basically all of the nickel from this solution is extracted at his point. Hence loading of nickel onto the organic phase is primarily a factor of the NaOH addition and not the initial nickel concentration used during pre-loading. From Figure 3-5 (b), higher initial nickel aqueous concentrations generates the same loading at lower pH values than lower aqueous concentrations do at higher pH values. A possible explanation is the fact that at the lower nickel concentrations the additional NaOH increases the solution pH once nickel in the aqueous phase becomes exhausted.

Figure 3-6 reveals what nickel loadings are required to extract cobalt and iron from the pregnant nickel solution at an O/A ratio of 1.

100 90

80 70 60 pre-loading with 6.15g/l Ni 50 pre-loading with 10.5g/l Ni 40 pre-loading with 21.2g/ Ni 30 Extraction (%) Extraction Co 20 Fe 10 0 0 1 2 3 4 5 6 7 8 9 10 11

Nickel in organic (g/l)

Figure 3-6: Extraction of cobalt (1 g/l) and iron (3 g/l) from the 80 g/l nickel solution as a function of nickel loading (see Table A-5 in Appendix A for the experimental data and conditions)

Stellenbosch University http://scholar.sun.ac.za Batch tests

Referring to Figure 3-6, more than 90% extraction is achieved at nickel loadings of 3 g/l and 6 g/l for iron and cobalt respectively. The higher nickel loading (6 g/l) is more than the required loading according to the stoichiometry in reaction 17 where 1 mole of Ni has to be loaded for each mode of impurity being extracted (nickel loading of 6 g/l (0.1 mol/l) Ni to remove a total of 4 g/l (0.07 mol/l) of Co and Fe). O'Callaghan and Chamberlain (2002) and Nogueira et al. (2009) found that a nickel loading of 7.5 g/l is adequate to remove more than 98% of 6.5 g/l Co, 11 mg/l Cu, 5 mg/l Zn, and 11 mg/l Mn from a 78.1 g/l nickel solution at an A/O ratio of 1. Nogueira et al. (2009) reported that using nickel loadings ranging between 7 and 13 g/l are sufficient to purify an 83 g/l nickel stream from 3 g/l of cobalt at an A/O ratio of 0.2. These two publications compares well with the data displayed in Figure 3-6.

From Figure 3-6 it is also clear that using different initial nickel concentrations of 6.15, 10.5 and 21.2 g/l during pre-loading does not have a severe effect on extraction and that the final nickel loading (irrespective of the initial nickel aqueous concentration used during pre-loading) of Cyanex 272 will determine cobalt and iron extraction.

Next the sodium additions to the purified nickel stream will be considered. Figure 3-7 shows the extraction curves generated when using pure Cyanex 272 without any nickel pre-loading as well as the sodium additions to the product solution after extraction as a function of pH.

100 7 90

80 6

70 Ni 5 60 4 50 Co phase aqueous 40 Fe 3

Extraction (%) Extraction 30 Na 2

20 (g/l) extraction after 1 10 0 0 to addition Na 2.5 3 3.5 4 4.5 5 5.5 6 6.5 pH

Figure 3-7 shows that more than 95% of cobalt and iron are extracted from the nickel solution at pH values of 3.5 and 5, respectively, which corresponds well to typical pH values from literature (Miralles et al., 1992; Hubicki and Hubicka, 1996, Park, 2007). The maximum nickel extraction is approximately 10% or 8 g/l at a pH of 6.5. At a pH of 5, approximately 4 g/l of sodium reports to the nickel product solution. Consequently, while 1 g/l cobalt and 3 g/l iron are being removed from the product solution, 4 g/l of sodium is added instead. Figure 3-8 shows the extraction curves generated when using the nickel salts of Cyanex 272 for cobalt and iron separation as well as the sodium addition to the nickel product solution as a function of pH.

Stellenbosch University http://scholar.sun.ac.za Batch tests

0.25 100

90 0.20 after

70 0.15 60 50 0.10

40 phase aqueous 0.05

Extraction (%) Extraction 30 extraction (g/l) extraction 20 0.00 10 0 -0.05

2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 to addition Na pH

pre-loading with 6.15g/l Ni pre-loading with 10.5g/l Ni pre-loading with 21.2g/l Ni Ni Co Fe Na

It is noted from Figures 3-6 and 3-8 that higher nickel loadings onto the organic phase results in higher equilibrium pH values. This is caused by the fact that when nickel is exhausted from the organic phase that hydrogen is exchanged for iron and then causes a drop in pH of the solution as discussed previously.

Cobalt and iron are extracted almost completely at pH values of 5 and 3.5, respectively, which correspond with the values obtained when using pure Cyanex 272. Maximum nickel extraction with Ni- Cyanex 272 is 5.18% compared to 12.61% with pure Cyanex 272. This is due to the fact that nickel is also released from the organic phase into the aqueous phase with Ni-Cyanex 272 and not only extracted. Very little sodium addition to the nickel product solution was observed. At a pH of 5 the removal of 1 g/l cobalt and 3 g/l iron from the product solution resulted in the addition of only 0.05 g/l sodium to the nickel product solution with Ni-Cyanex 272 compared to 4 g/l with pure Cyanex 272. It is therefore concluded that the pre-loading procedure is the best alternative based on the low sodium additions to the purified product solution.

3.3.1.3 EFFECT OF VARIOUS DILUENTS

Figure 3-9 and 3-10 shows the results for pre-loading Cyanex 272 and extraction with the pre-loaded Cyanex 272, respectively, by using various diluents. All the diluents gave good phase separation at 50°C which meant that no third phase modifier was necessary.

Stellenbosch University http://scholar.sun.ac.za Batch tests

9 9 8 8

7 7 6 6 5 5 4 4 organic (g/l) organic 3 (g/l) organic 3 - -

Ni 2 Ni 2 1 1 0 0 5.3 5.8 6.3 6.8 7.3 0.0 0.5 1.0 1.5 2.0 pH NaOH/Nio (mol/mol) (a) (b) 9 9 8 8

7 7 6 6 5 5 4 4

organic (g/l) organic 3 organic (g/l) organic 3 - -

Ni 2 Ni 2 1 1 0 0 5.3 5.8 6.3 6.8 7.3 0.0 0.5 1.0 1.5 2.0 pH NaOH/Nio (mol/mol) (c) (d)

Kerosene Heptane Toluene pH Xylene Shellsol D70 0.54M Cyanex 272

Referring to Figure 3-9, all the diluents behaved very similar as a function of NaOH addition. In contrast they do however have an effect on the pH at which nickel extraction occurs as shown in Figures 3-9 (a) and (c). When considering Xylene and Heptane for example in Figures 3-9 (a) and (c), nickel loading differs approximately with 1 g/l and 2 g/l, respectively, at a pH of 6.3.

The loading curves in Figure 3-9 (d) (variance at NaOH/Nio values of 0.21, 0.43, 0.85 and 1.71 are 0.12, 0.02, 0.07 and 0.02, respectively) are closer to each other than in Figure 3-9 (b) (variance at NaOH/Nio values of 0.21, 0.43, 0.85 and 1.71 are 0.18, 0.19, 0.14 and 0.11 respectively). A possible explanation is the fact that a higher concentration of Cyanex 272 is used in Figure 3-9 (d) (20V%) than in (b) (10V%) which means that the interactions between the diluent and ligand are less significant in the former than the latter case caused by the excess extractant.

Extraction of nickel, cobalt and iron with Ni-Cyanex 272 in Figure 3-10 is considered next.

Stellenbosch University http://scholar.sun.ac.za Batch tests

4 4 2 2 0

0 -2 -2 -4 -4 -6 -6 -8 -8 Ni extraction(%) Ni -10 extraction(%) Ni -10 -12 -12 -14 -14 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9

Ni - organic (g/l) Ni - organic (g/l) (a) (b)

100 100 90 90

80 80

70 70 60 60 50 50 40 40 30 30 Co extraction(%) Co 20 extraction(%) Co 20 10 10 0 0 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 Ni - organic (g/l) Ni - organic (g/l) (c) (d)

100 90 100 90 80

80 70 70 60 60 50 50 40 40 30

Fe extraction(%) Fe 30 20 extraction(%) Fe 20 10 10 0 0 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 Ni - organic (g/l) Ni - organic (g/l) (e) (f)

Kerosene Heptane Toluene pH Xylene Shellsol D70 0.54M Cyanex 272

Stellenbosch University http://scholar.sun.ac.za Batch tests

For nickel extraction with Ni-Cyanex 272 the data is once again scattered (Figures 3-10 (a) and (b)) as was the case with the kinetic data (the negative extraction is due to the fact that more nickel is released from the organic than extracted from the aqueous phase). For all the diluents nickel extraction however remained below 4% which means that less than 3.28 g/l of nickel was lost from the 81.97 g/l solution, irrespective of the loading or diluent used. From Figures 3-10 (c)-(f) the diluents influence on the extracting power of Cyanex 272 for cobalt and iron once again becomes less severe when using a 20V% rather than a 10V% solution for the same reasons as discussed previously for pre-loading.

Since the various diluents do not have a significant effect on the extracting behaviour of Cyanex 272, Shellsol D70 was chosen as diluent for all subsequent experimental work given the fact that it is currently being used on industrial scale for the extraction of cobalt and nickel with Cyanex 272 (Flett, 2005).

3.3.1.4 FACTORIAL DESIGNS

The pre-loading and extraction factorial designs will be discussed in this section.

3.3.1.4.1 Pre-loading

The repeatability of the pre-loading setup will be considered prior to interpretation of the factorial design results. Equation 21 was used to define repeatability (Measey et al., 2003):

[21]

where: MSbetween = Mean squares between groups (between run 1-16)

MSwithin = Mean squares within groups (within each run/3 repeats) n = Number of repeated measurements r = Repeatability

Repeatability varies from 0 to 1 and expresses the fraction of variation that is due to differences between groups not due to differences within groups. If the average group is consistent, then the average within group variation will be low which means that the ratio of among group variation to within group variation (the repeatability) will be high.

Table 3-3 gives qualitative terms to describe repeatability as calculated with equation 21.

Stellenbosch University http://scholar.sun.ac.za Batch tests

Table 3-3: Terms to express the repeatability values calculated with equation 21 (adapted from (Measey et al., 2003))

r less than 0.2 Slight repeatability r between 0.2 and 0.4 Low repeatability r between 0.4 and 0.7 Moderate repeatability r between 0.7 and 0.9 High repeatability r greater than 0.9 Very high repeatability

A single factor ANOVA on the pre-loading factorial design yielded the mean square (MS) values between and within groups for nickel loading, the sodium concentration in the aqueous solution after extraction and pH values. Equation 21 was subsequently used to calculate the repeatability, which is displayed in Table 3-4.

Table 3-4: Repeatability (r) values for the pre-loading factorial design

Ni loading Generated Na stream pH MS n r MS n r MS n r Between Groups 114.15 5.20 1.09 3 0.996 3 0.982 3 0.997 Within Groups 0.17 0.03 0.0009

According to the r-values in Table 3-4, pH gives the best repeatability followed by the nickel loading and then the sodium addition to the aqueous phase after extraction with r values of 0.997, 0.996 and 0.982 respectively. In all three instances, repeatability is considered as very high according to Table 3-3.

Two regression models were fitted to generated data from the pre-loading factorial design as shown in Table 3-5. The p-values for a 95% confidence interval for the two models are also shown.

Table 3-5: P-values for the pre-loading 24 factorial design with the high and low values shown in Table 3-1

y = zo + z1a + z2b + z3c + z4d + z5ab + y = zo + z1a + z2b + z3c + z4d z6ac + z7ad + z8bc + z9bd + z10cd Na Na addition Ni addition Ni loading after Variable loading after pH pH (g/l) extraction (g/l) extraction (g/l) (g/l) A/O ratio (a) 6.E-22 0.23 5.E-24 a 2.E-50 0.22 4.E-31 Extractant concentration (b) 0.68 0.23 6.E-15 b 0.01 0.22 2.E-22 NaOH addition (c) 3.E-13 2.E-34 6.E-24 c 3.E-41 2.E-31 4.E-31 Temperature (d) 0.90 0.14 0.08 d 0.39 0.13 3.E-03 ab 1.E-03 0.15 1.E-06

ac 8.E-34 0.90 0.13

ad 0.69 0.16 6.E-06

bc 0.90 0.55 0.14

bd 0.28 0.92 9.E-05

cd 0.62 0.05 5.E-05

Stellenbosch University http://scholar.sun.ac.za Batch tests

It is clear that A/O ratio and NaOH addition are the two dominating factors with respect to nickel loading with p-values of 6x10-22 and 3x10-13 respectively. Extractant concentration and temperature on the other hand do not affect nickel loading to such a large extent with p-values of 0.68 and 0.9 respectively. The following preliminary experiments confirm this statement.

10 10 9 9 8 8 7 7 6 6 5 5 4

4 organic (g/l) - organic (g/l) organic - 3

3 Ni Ni 2 2 1 1 0 0 0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3 Volume (ml) of 3M NaOH for O/A of Volume (ml) of 3M NaOH for O/A of 20ml/20ml 20ml/20ml (a) (b) 25ºC (Nio = 6g/l) 10V% Cyanex 272 (Nio = 6.15g/l) 25ºC (Nio = 10g/l) 10V% Cyanex 272 (Nio = 10.5g/l) 50ºC (Nio = 6g/l) 20V% Cyanex 272 (Nio = 6.15g/l) 50ºC (Nio = 10g/l) 20V% Cyanex 272 (Nio = 10.5g/l)

Figure 3-11: (a) Influence of temperature on nickel loading as a function of NaOH addition (b) Influence of extractant concentration on nickel loading as a function of NaOH addition (see Tables A-12 and A-13 in Appendix A for the experimental data and conditions)

From Figure 3-11 (a), nickel loading increases by approximately 1 g/l as temperature increases from 25°C to 50°C for 3 M NaOH additions higher than 1.25 ml. This has been confirmed by other authors as discussed during the literature review in section 2.7.3 (Bhaskara Sarma and Reddy, 2002; Sarangi et al., 1999b). It was concluded that the effect of temperature on nickel loading is not as severe as O/A ratio and NaOH addition and all subsequent pre-loading tests were conducted at 50°C. Figure 3-11(b) confirms the relatively high p-value (0.68) for extractant concentration with respect to nickel loading in Table 3-5. Therefore only the influence of NaOH addition and A/O ratio on nickel loading was further investigated with respect to nickel loading.

Referring to Table 3-5 the Na addition to the aqueous solution after extraction is only influenced by the NaOH addition to the mixture. This means that most of the added sodium will exit with the aqueous raffinate stream used for pre-loading. All the variables except temperature have a significant influence on the pH. Most authors publish their results as a function of pH and not the quantity of NaOH added. Gandhi et al, (1993) for instance reported that extraction increases as the extractant concentration increases since extraction then occurs at a lower pH. This is not entirely true since the extractant concentration has an effect on the equilibrium pH and not extraction. Figure 3-12 illustrates this statement by plotting the same results from Figure 3-11(b) but only this time as a function of pH and pH as a function of NaOH addition.

Stellenbosch University http://scholar.sun.ac.za Batch tests

10 9.0 9 8 8.5 7 8.0 6 7.5

5 7.0 4 pH 6.5 organic (g/l) organic

- 3 6.0 Ni 2 1 5.5 0 5.0 5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5 9.0 0 0.5 1 1.5 2 2.5 pH Volume (ml) of 3M NaOH for A/O 20ml/20ml (a) (b) 100 10V% Cyanex 272 (Nio = 6.15g/l) 10V% Cyanex 272 (Nio = 10.5g/l) - c ni or ga Ni 5.020V% Cyanex5.5 272 (Nio6.0 = 6.15g/l)6.5 7.0 7.520V% Cyanex8.0 272 (Nio8.5 = 10.5g/l)9.0 pH …

Figure 3-12: Influence of extractant concentration on nickel loading as a function of pH (see Tables A-12 and A-13 in Appendix A for the experimental data and conditions)

Referring to Figure 3-12 (a), higher extractant concentrations extract nickel to the same extent at lower pH values than lower extractant concentrations do at a higher pH values. Hence using more extractant decreases the pH at which extraction occurs caused by the extra protons being added to the system coming from the phosphinic acid (Cyanex 272) as shown in Figure 3-12 (b). This is not the case in Figure 3-11(b) where both extractant concentrations extract nickel to the same extent as a function of NaOH addition.

3.3.1.4.2 Extraction

Next the repeatability of the extraction runs will be considered followed by a discussion of the results obtained from the factorial design. The repeatability as calculated with equation 21 for nickel, cobalt and iron extraction as well as the generated equilibrium pH are shown in Table 3-6.

Table 3-6: Repeatability (r) values for the extraction factorial design

Nickel extraction Cobalt extraction Iron extraction pH MS n r MS n r MS n r MS n r Between Groups 38.13 5675.46 1855.02 2.25 3 0.371 3 0.996 3 0.988 3 0.990 Within Groups 13.76 7.60 7.27 0.008

According to Table 3-3, cobalt extraction, iron extraction and pH are considered as very highly repeatable with r-values very close to 1. Nickel extraction, however, is not repeatable according to the r- value of 0.371. Referring to equation 21 this can be explained by the fact that the mean square between groups (38.13) does not differ as much from the mean square within groups (13.76) as with cobalt, iron and pH. This was expected since nickel extraction rarely exceeded 10% resulting in the low MS (between groups) value of 38.13 compared to 5675.46 and 1855.02 for cobalt and iron. The MS value within groups of 13.76 for nickel compares better with the values of 7.6 and 7.27 for cobalt and iron. This, however, is an indication that the nickel extraction data remained relatively consistent and that the poor repeatability is a result of the low variance there was between the data groups.

Stellenbosch University http://scholar.sun.ac.za Batch tests

Two regression models were fitted to the extraction tests. These models and the p-values for a 95% confidence interval for the extraction factorial design are shown in Table 3-7.

Table 3-7: P-values for the extraction 2 4 factorial design with the high and low values shown in Table 3-1

y = zo + z1a + z2b + z3c + z4d + z5ab + y = zo + z1a + z2b + z3c + z4d z6ac + z7ad + z8bc + z9bd + z10cd Extraction (%) Extraction (%)

Variable Ni Co Fe pH Ni Co Fe pH

O/A ratio (a) 0.026 5.E-40 3.E-20 4.E-24 a 0.039 1.E-44 2.E-36 1.E-28 Extractant concentration (b) 0.130 0.572 0.604 2.E-06 b 0.283 0.330 0.103 4.E-10 Ni loading (c) 3.E-05 3.E-06 5.E-07 2.E-12 c 0.003 2.E-11 1.E-20 3.E-17 Temperature (d) 0.004 0.006 0.030 2.E-06 d 0.661 1.E-05 2.E-08 4.E-10 ab 0.042 0.271 0.104 0.034

ac 0.647 3.E-09 2.E-20 7.E-04

ad 0.001 2.E-06 3.E-08 4.E-07

bc 0.049 0.764 0.932 0.022

bd 0.773 0.379 0.128 0.225

cd 0.02 0.910 0.613 0.304

Due to the poor repeatability of the nickel extraction data it is difficult to come to any reasonable conclusions regarding its behaviour as a function of the design variables. The main objective is however to remove cobalt and iron from the pregnant nickel solution without losing too much nickel in the process. Hence predictions will only be made of the maximum and minimum nickel losses that can be expected in succeeding experiments.

O/A ratio with p-values of 5x10-40 and 3x10-20 and nickel loading with p-values of 3x10-6 and 5x10-7 are the dominating factors with respect to cobalt and iron extraction, respectively. Consequently these two variables were considered during the final experiments to determine the operating conditions for the extraction section. Extractant concentration once again does not appear to have a significant influence on cobalt and iron extraction according to their relatively high p-values (0.572 and 0.604). Figure 3-13 displays preliminary runs that were conducted to investigate the influence of extractant concentration.

100 100

90 90 80 80 70 70 60 60 50 50 40 40 30 30 20 20 Fe extraction (%) extraction Fe Co extraction (%)extraction Co 10 10 0 0 0 2 4 6 8 10 12 0 2 4 6 8 10 12 Ni - organic (g/l) Ni - organic (g/l) (a) (b) 10V% Cyanex 272 Ni - organic20V% (g/l) Cyanex 272 50ºC

Figure 3-13: Effect of extractant concentration on cobalt and iron extraction with Ni-Cyanex 272 (see Table A-14 in Appendix A for the experimental data and conditions)

Stellenbosch University http://scholar.sun.ac.za Batch tests

Based on Figure 3-13 (a) the use of a higher extractant concentration does in fact increase cobalt extraction for nickel loadings higher than 4.3 g/l which contradicts the results in Table 3-7. The maximum nickel loading that was used during the factorial design is 4.76 g/l (see Table 3-1). This is close to 4.3 g/l and with the O/A ratio being varied as well during the factorial design the effect of extractant concentration on cobalt extraction for higher nickel loadings was not revealed by the statistics. Extractant concentration does however, not have a significant effect on iron extraction as shown in Figure 3-13(b). It is well published in literature that cobalt extraction increases as the extractant concentration increases (Sarangi et al., 1999b; Swain et al., 2006, Park, 2007). Consequently it was concluded that a 20V% solution of Cyanex 272 will be used due to the higher extraction obtained for cobalt as well as for the fact that it reduces the probability of phase separation issues caused by overloading of the ligand. As the Cyanex 272 concentration increases, the organic phase viscosity increases. Typically Cyanex 272 concentrations of not higher than 20V% are used since the organic phase then becomes too viscous.

According to Table 3-7 temperature influences cobalt and iron extraction with p-values of 0.006 and 0.03 respectively. Figure 3-14 shows preliminary results that will clarify some of the influences that temperature may have on cobalt and iron extraction.

100 100 90 90

80 80 70 70 60 60 50 50 40 40 30 30 Fe extraction (%) extraction Fe

Co extraction (%)extraction Co 20 20 10 10 0 0 0 2 4 6 8 10 12 0 2 4 6 8 10 12 Ni - organic (g/l) Ni - organic (g/l) (a) (b) 100 0 25ºC 50ºC

Figure 3-14: Effect of temperature on cobalt and iron extraction with Ni-Cyanex 272 (see Table A-15 in Appendix A for the experimental data and conditions)

Referring to Figure 3-14 (a), cobalt extraction increases as temperature decrease for nickel loadings less than approximately 4 g/l. Hence when there is not enough nickel present in the organic phase temperature does in fact influence cobalt extraction. Sarangi et al. (1999b) also observed that cobalt extraction decreases as temperature increases within the range 30-50°C as shown in Figure 2-10. According to Figure 3-14(b) temperature has no significant influence on iron extraction. Consequently the influence of temperature was not further considered since phase separation issues were observed at temperatures below 30°C (see section 2.5) and there will be sufficient nickel in the organic phase to ensure complete removal of cobalt and iron from the pregnant nickel solution. A temperature of 50°C was chosen for all the subsequent extraction experiments.

Stellenbosch University http://scholar.sun.ac.za Batch tests

The interaction between O/A ratio and nickel loading for cobalt and iron extraction in Table 3-7 can be explained by the fact that cobalt and iron extraction are largely dependent on the quantity of nickel in the organic phase rather the concentration/loading. Hence a higher loading will be sufficient at a low O/A ratio and a lower loading at a higher O/A ratio. The A/O ratio and nickel loading should just be monitored to prevent overloading to greater than 60% of the ligand resulting in phase separation problems.

The equilibrium pH is influenced by all the variables (refer to Table 3-7). The O/A ratio and nickel loading influence can once again be explained by the fact that hydrogen will be exchanged for iron once nickel is exhausted from the organic phase when the pH is high enough. The influence of extractant concentration is due to the extra hydrogen atoms being added as discussed previously and temperature typically has an influence on pH as explained by the Nernst equation.

3.3.1.5 O/A RATIO , NICKEL LOADING AND NAOH ADD IT ION

The nickel loading and O/A ratio requirements for the extraction section were studied prior to the pre- loading section since it is important to first determine the loading requirements for the extraction section. Figure 3-15 reveals what typical nickel losses and sodium additions to the purified nickel stream might be expected after extraction.

10 0.20

6 0.15

4 0.10 2

0 0.05 -2 Ni extraction (%) extraction Ni

-4 0.00 -6 Na addition after addition (g/l) extraction Na -8 -0.05 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0

O/A (ml/ml) O/A (ml/ml) (a) (b)

3.0 Ni-organic = 0.56g/l Ni-organic = 1.97g/l Ni-organic = 2.94g/l

pH 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Ni-organic = 3.95g/l Ni-organicO/A (ml/ml) = 4.69g/l Ni-organic = 5.73g/l

Stellenbosch University http://scholar.sun.ac.za Batch tests

The negative extraction of nickel in Figure 3-15(a) is caused by the fact that more nickel is transferred from the organic to the aqueous phase than from the aqueous to the organic phase. The nickel extraction data are once again relatively scattered and ranges between -6 and 6 percent with the exception of 8.77% when using a nickel loading of 0.56 g/l at an O/A ratio of 2. Apart from this exception maximum nickel losses of approximately 4.8 g/l can be expected. Referring to Figure 3-15(b) the sodium additions to the purified nickel stream predominantly increases as the nickel loading increases for O/A ratios higher than 1 (The negative sodium additions are due to impure sodium that was in the salts used to prepare the synthetic representation of the pregnant nickel leach solution which was then also extracted and typically ranges between 0.05-0.1 g/l). This can be explained by the fact that higher nickel loadings during the pre-loading stage demands higher NaOH additions and consequently increases the probability of sodium uptake by the extractant causing the higher sodium additions to the purified pregnant nickel solution. Nonetheless typical maximum sodium additions of 0.1 g/l might be expected during the extraction section with the exception of 0.17 g/l for a high nickel loading of 5.73 g/l and an O/A ratio of 2.5.

Next cobalt and iron separation will be considered as a function of O/A ratio and nickel loading.

100 1.4 90 80 1.2

70 1.0 60 0.8 50 organic (g/l) organic

40 - 0.6

30 Co Co extraction (%)extraction Co 0.4 20 0.2 10 0 0.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Co-aqueous (g/l) O/A (ml/ml) (b) (a)

100 6 90 80 5

70 4 60 50 3

40 (g/l) organic - 2 30 Fe

Fe extraction (%) extraction Fe 20 1 10 0 0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Fe-aqueous (g/l) O/A (ml/ml) (d) (c)

Stellenbosch University http://scholar.sun.ac.za Batch tests

70 5.5 60

5.0 50

4.5 40 pH 30 4.0

E(%)Fe/E(%)Co E(%)Fe/E(%)Co 20 3.5 10

3.0 0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 O/A (ml/ml) O/A (ml/ml) (e) (f)

Ni-organic = 0.56g/l Ni-organic = 1.97g/l Ni-organic = 2.94g/l

3.0 O/A (ml/ml) pH Ni-organic = 3.95g/l Ni-organic = 4.69g/l Ni-organic = 5.73g/l 0.0 0.5 1.0 1.5 2.0 2.5 3.0

Figure 3-16: Variation in O/A ratio and nickel loading for pH and cobalt (1 g/l) and iron (3 g/l) extraction (see Table A-16 in Appendix A for the experimental data and conditions)

Referring to Figure 3-16 (a) and (c), higher nickel loadings reduce the O/A ratio at which cobalt and iron extraction occurs which confirms the earlier statement in section 3.3.1.4.2 that extraction is largely a function of the nickel content in the organic phase rather than the nickel concentration. This is also reflected in Figure 3-16(e) where higher nickel loadings, or rather the nickel content in the organic phase, increase the equilibrium pH.

Most frequently extraction McCabe-Thiele diagrams are reported as a function of pH. This means that NaOH is added within each stage to control the solution pH. This approach, however, is limited when using the pre-loading procedure since the McCabe-Thiele diagrams are generated at various nickel loadings. It will be more challenging to control the nickel loading between extraction stages than having to control the pH (to do this a separate pre-loading section will be required within each extraction stage). The McCabe-Thiele diagrams (Figures 3-16 (b) and (d)) will therefore only be used to confirm that the selected nickel loading and O/A ratio are sufficient to extract cobalt and iron in a single stage.

Referring to Figure 3-16, two different scenarios will be considered. In the first one the selective extraction of iron and cobalt will be discussed followed by a discussion of the simultaneous extraction of cobalt and iron.

Referring to Figure 3-16 (f), a plot is given of iron extraction divided by cobalt extraction (E(%)Fe/E(%)Co) as a function of nickel loading and O/A ratio. The aim of this plot is to identify a nickel loading and O/A ratio that can selectively separate iron and cobalt from the pregnant solution. The higher the value for

(E(%)Fe/E(%)Co) the better the selectivity. The highest value for E(%)Fe/E(%)Co is 64.4 at an O/A ratio of 0.25 and a nickel loading of 4.69 g/l. At these conditions cobalt extraction is very low (0.63%) but only 40.6% iron extraction is reached, referring to Figures 3-16 (a) and (c), respectively. The second highest value obtained for (E(%)Fe/E(%)Co) is 52 at an O/A ratio of 0.5 and a nickel loading of 3.95 g/l. 75% iron extraction is attained at these conditions as shown in Figure 3-16(c). The third best value for

Stellenbosch University http://scholar.sun.ac.za Batch tests

(E(%)Fe/E(%)Co) is 23 at an O/A ratio of 0.25 and a nickel loading of 5.73 g/l with iron being extracted up to 50%. Referring to Figure 3-16 (f), the highest values for E(%)Fe/E(%)Co are reached at O/A ratios less than 0.5. Even though very little cobalt extraction occurs at these conditions, phase separation issues were encountered caused by ligand overloading (see section 2.5). Ultimately the best operating conditions with respect to selectivity would be at an O/A ratio of 1.5 and a nickel loading of 1.97 g/l giving 7.25% cobalt and 98.79% iron extraction resulting in no phase separation problems.

For the second scenario, only cobalt extraction has to be considered since sufficient iron extraction is expected at these conditions. Referring to Figures 3-16 (a) and (c), an O/A ratio of 1.5 with a nickel loading of 3.2 g/l is selected since it gives maximum cobalt (95-96%) and iron (99-100%) extraction (this loading was selected via interpolation). This is a high enough O/A ratio which reduces possible ligand overloading and the nickel loading is relatively low compared to the other higher loadings which minimizes the quantity of nickel that has to be recycled from the purified product solution for pre- loading. From the McCabe-Thiele diagrams in Figures 3-16 (b) and (d), a single stage is sufficient to extract cobalt and iron at this O/A ratio and nickel loading.

The primary objective of the project is to purify the concentrated nickel solution while selective separation of cobalt and iron are secondary. Due to the unpredictability of the nickel extraction data, a pure cobalt and iron product cannot be guaranteed with the selective extraction approach and the second scenario was selected as the best alternative. This also eliminates the additional pre-loading unit that will be required (one before iron extraction and one before cobalt extraction). Selective stripping of cobalt, iron and any nickel that may still be left in the organic phase was subsequently considered.

From Figure 3-16 (b), cobalt is still present in quantities ranging between 0.03-0.05 g/l after extraction for the highest nickel loadings and O/A ratios. For complete cobalt extraction minor additions of NaOH to the extractor unit will ensure complete cobalt removal as well since this will slightly increase the pH. An approximation of the additional sodium that will end up in the nickel product solution to remove a remaining 0.05 g/l of cobalt is 0.04 g/l [(0.05/59)(2)(23) = (concentration of cobalt left after extraction/atomic weight of cobalt)(moles of sodium required to neutralize the total moles of H+ being released for cobalt)(atomic weight of sodium)].

Next the O/A ratio and pH for the pre-loading section will be considered. If a portion of the purified nickel stream is used (80 g/l) to pre-load Cyanex 272 (refer to Figure 2-5), then the corresponding O/A ratio to reach these loadings (assuming complete nickel extraction at the pre-loading unit) is determined via the following equation which is derived via a mass-balance:

[22]

where O/A = Organic/aqueous mixing ratio [ml/ml]

Vorg; Vaq = Volumes of the organic and aqueous phase [g/l]

Niorg; Niaq = Nickel concentration of the organic and aqueous phases [g/l]

Stellenbosch University http://scholar.sun.ac.za Batch tests

The aim was to generate nickel loadings ranging between 3-6 g/l with an 80 g/l nickel aqueous solution. The reason for using such a high nickel concentration was to eliminate unnecessary water additions to the system to dilute the purified nickel stream used for pre-loading. If 3 g/l and 6 g/l are substituted in equation 22 the O/A ratio interval generated is 13.33-26.667. The pre-loading results as functions of the O/A ratio and pH are shown in Figure 3-17:

100 7.0 95 6.5 90 6.0

85 5.5 80 5.0 75 4.5 70 4.0

65 (g/l) organic - 3.5

60 Ni 3.0 Ni extraction(%) Ni 55 50 2.5 45 2.0 5.5 5.7 5.9 6.1 6.3 6.5 6.7 6.9 7.1 7.3 7.5 5.5 5.7 5.9 6.1 6.3 6.5 6.7 6.9 7.1 7.3 7.5 pH pH (a) (b) 2.07.0

- O/A = 26.67 O/A = 20 O/A = 16 O/A = 13 r g a n i o N 5.5 5.7 5.9 6.1 6.3 6.5 6.7 6.9 7.1 7.3 7.5 7 pH …

3 organic (g/l) organic - 2 pH = 6.9-7 Ni pH = 6.4-6.7 1 pH = 5.8-6 0 0 10 20 30 40 50 60 70 80 Ni-aqueous (g/l) Stage 2 for pH (c) between 6.4-6.4

Figure 3-17: Variation in O/A ratio and pH with respect to nickel loading of Cyanex 272 (see Table A-17 in Appendix A for the experimental data and conditions)

Referring to Figure 3-17 (a), at pH values higher than 6.7 the O/A ratio does not have a significant influence on extraction and at a pH higher than 7.1 all of the nickel is extracted irrespective of the O/A ratio. As expected the nickel loading decreases when using higher O/A ratios as shown in Figure 3-17 (b). To remove cobalt and iron from the pregnant solution a nickel loading of 3.2 g/l is required as discussed previously: substituting this in equation 22 results in an O/A ratio of 25. From Figure 3-17(a) a pH of approximately 7 will be sufficient to completely extract nickel at this O/A ratio and generate the required loading in a single stage as shown in Figure 3-17(c). Decreasing the pH to values ranging between 6.4 and 6.7 will require an additional stage to extract the remaining nickel as indicated in Figure 3-17 (c). The higher pH of 7 is, however, proposed since a single pre-loading unit will then be required instead of two units.

Stellenbosch University http://scholar.sun.ac.za Batch tests

3.3.2 STRIPPING

Nickel was also loaded onto the organic phase along with cobalt and iron during the stripping tests. The aim was to see whether selective stripping of nickel, cobalt and iron is possible. The stripped nickel can then be rejoined with the purified pregnant solution while pure cobalt and iron are recovered from the organic phase.

3.3.2.1 KINETICS

The results for the kinetic tests with respect to stripping are shown in Figure 3-18.

100 90 80 70 60 Ni (0.1M H₂SO₄) 50 Ni (0.5M H₂SO₄) 40 Co (0.1M H₂SO₄) Co (0.5M H₂SO₄)

Stripped (%) Stripped 30 Fe (0.1M H₂SO₄) 20 Fe (0.5M H₂SO₄) 10 0 0 100 200 300 400 500 600 700 800 900 Agitation time (s)

Figure 3-18: Stripping kinetics of an organic loaded with nickel (4.5 g/l), cobalt (1.04 g/l) and iron (2.96 g/l) (see Table A-18 in Appendix A for the experimental data and conditions)

Form Figure 3-18 it is clear that the stripping kinetics are slower than the extraction kinetics with Ni- Cyanex 272 with equilibrium being reached after approximately 410 seconds compared to 150 seconds (refer to Figure 3-4). An agitation period of 600 seconds was therefore allowed for succeeding stripping tests.

3.3.2.2 FACTORIAL DESIGN

The repeatability that was calculated with equation 21 for the factorial design with respect to percentage nickel, cobalt and iron stripped as well as the equilibrium pH is shown in Table 3-8.

Table 3-8: Repeatability (r) values for the stripping factorial design

Nickel stripping Cobalt stripping Iron stripping pH MS n r MS n r MS n r MS n r Between Groups 595.56 2996.12 3683.11 5.32 3 0.962 3 0.993 3 0.996 3 0.997 Within Groups 7.71 7.52 4.34 0.006

Stellenbosch University http://scholar.sun.ac.za Batch tests

All the r-values in Table 3-8 are very close to 1 which means that the stripping tests are also highly repeatable according to Table 3-3.

The p-values for the two regression models fitted to the stripping factorial design data, as shown in Table 3-9 will be considered next.

Table 3-9: P-values for the stripping 2 4 factorial design with the high and low values shown in Table 3-2

y = zo + z1a + z2b + z3c + z4d + z5ab + y = zo + z1a + z2b + z3c + z4d z6ac + z7ad + z8bc + z9bd + z10cd Stripped (%) Stripped (%) Variable Ni Co Fe Ni Co Fe A/O ratio (a) 3.E-04 2.E-03 2.E-06 a 1.E-08 7.E-07 1.E-13 Acid concentration (b) 2.E-04 4.E-04 9.E-22 b 6.E-09 3.E-08 3.E-31 Metal loading (c) 7.E-06 6.E-05 0.08 c 2.E-11 1.E-09 6.E-04 Temperature (d) 7.E-03 0.399 0.911 d 6.E-06 0.13 0.82 ab 1.E-06 3.E-06 1.E-11 ac 0.01 1.E-05 5.E-04 ad 0.117 0.458 2.E-05 bc 9.E-10 4.E-08 3.E-03 bd 0.8 0.825 0.07 cd 0.274 0.511 0.05

Referring to Table 3-9, temperature only influences nickel stripping. Nickel stripping, however, only increased with values ranging between 5-10% as the temperature increased. This effect was considered as being not large enough and a temperature of 50°C was chosen for all subsequent stripping experiments for the same reasons as the pre-loading and extraction runs.

A high and low metal loading were chosen to see if the metal content in the organic phase influences the percentage of metal stripped or if the acid concentration and A/O ratio are the two dominating factors. This is important since the O/A ratio used during the extraction section will determine the metal concentration in the organic phase and possibly influence the extent of stripping. From Table 3-9 it is clear that the metal concentration does, in fact, influence the percentage of metal stripped, with p- values well below 0.05 except for iron with a p-value of 0.08. This means that the metal loadings that will be used for the remaining stripping tests will be based on the O/A ratio of the extraction section.

Interactions also exist between the metal loading with A/O ratio and the acid concentration. The only exception is for nickel where there is not a significant interaction between the metal loading and A/O ratio. A possible explanation is the fact that nickel is stripped more easily than cobalt or iron and the acid concentrations used were high enough to ensure that cobalt and iron striping also occurred. Hence there was sufficient H+ for nickel to be stripped irrespective of the A/O ratio used. Stripping is therefore primarily dependent on the quantity of H+ cations available in the strip solution, i.e., the quantity of metal being removed from the organic phase is determined by the H+ content in the aqueous phase rather than the concentration of H+ in the aqueous phase.

Stellenbosch University http://scholar.sun.ac.za Batch tests

As expected the A/O ratio and acid concentration are the two dominating factors with respect to stripping; hence these two variables were further investigated to determine operating conditions to regenerate the organic phase as well as possibilities for selective stripping.

3.3.2.3 A/O RATIO AND ACID CONCENTRATION

Next the A/O ratio and H2SO4 concentration were considered. Table 3-10 indicates where phase separation issues (a single emulsion formed) were encountered as well as the relative concentrations at which nickel, cobalt and iron were stripped.

Table 3-10: Phase separation and selectivity defined by the relative quantities stripped of each metal

H2SO4 A/O (M) 0.50 0.75 1.00 1.50 2.25 3.00 Poor Poor Poor Poor Poor Poor 0.025 43 Inf Inf 85 Inf Inf 86 Inf Inf inf Inf Inf 262 Inf Inf 1.7 Inf Inf Poor Poor Poor Good Good Good 0.05 277 Inf Inf 160 Inf Inf 1.7 Inf Inf 1 Inf Inf 1 Inf Inf 1 Inf Inf Poor Poor Poor Good Good Good 0.075 332 Inf Inf 1.5 Inf Inf 1 Inf Inf 1 Inf Inf 1.2 Inf Inf 1 Inf Inf Poor Poor Poor Good Good Good 0.1 78 Inf Inf 1 Inf Inf 1 Inf Inf 1 Inf Inf 1.2 Inf Inf 1 Inf Inf Poor Poor Good Good Good Good 0.25 1 602 645 1 3.3 3.5 1 2.3 2.4 1 1.9 2 1.1 1.9 2.1 1 2 2.3 Good Good Good Good Good Good 0.5 1 1.4 1.5 1 1.2 1.2 1 1.1 1.2 1 1.2 1.2 1 1.2 1.2 1 1.2 1.3 Good Good Good Good Good Good 0.75 1 1.2 1.3 1 1.1 1.1 1 1.1 1.1 1 1 1.2 1 1.2 1.2 1 1.3 1.2 Good Good Good Good Good Good 1 1 1.3 1.4 1 0.9 0.9 1 1 1 1 1 1 1 1 1 1 1 1 Good Good Good Good Good Good 2 1 1 1.1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Good Good Good Good Good Good 4 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

Phase separation *Legend for interpretation of Table 3-10 S(%)Ni/S(%)Co S(%)Co/S(%)Fe S(%)Ni/S(%)Fe

The nickel, cobalt and iron stripped as well as pH as a function of acid concentration and A/O ratio are revealed in Figure 3-19.

Stellenbosch University http://scholar.sun.ac.za Batch tests

100 100 90 90

80 80 70 70 60 60 50 50 40 40 Ni stripped (%) stripped Ni 30 stripped (%) Co 30 20 20 10 10 0 0 0.00 0.05 0.10 0.15 0.20 0.25 0.00 0.05 0.10 0.15 0.20 0.25 H₂SO₄ (M) H₂SO₄ (M) (a) (b)

6 100 90 5

80 70 4 60

50 pH 3 40

Fe stripped (%) stripped Fe 30 2 20 10 0 1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.00 0.05 0.10 0.15 0.20 0.25 H₂SO₄ (M) H₂SO₄ (M) (c) (d)

A/O = 0.5 1001020304050607080900 A/O = 0.75 A/O = 1 H₂SO₄A/O = (M)1.5 A/O = 2.25 A/O = 3 …

i e p p ri N st d

Figure 3-19: Nickel (2.69 g/l), cobalt (0.75 g/l) and iron (2.08 g/l) stripping as a function of H 2SO4 concentration and A/O ratio (see Table A-21 in Appendix A for the experimental data and conditions)

Lower cobalt and iron loadings were used than those initially present in the pregnant solution since an O/A ratio of 1.5 was decided on for the extraction section. Ideally this would yield an organic phase with a loading of 0.7 g/l (1/1.5) Co and 2 g/l (3/1.5) Fe. Slightly higher concentrations were generated to rather overdesign the operating variables. 2.69 g/l of nickel was also loaded onto the organic phase to investigate selective removal of any co-extracted nickel as well.

From Figure 3-19 it is clear that stripping increases as the acid concentration and A/O ratio increases, with nickel being stripped initially followed by cobalt and then iron. First selective stripping of nickel and cobalt will be considered. Inf in Table 3-10 refers to infinity and was used when no stripping occurred for the component in the denominator as shown in the legend (i.e., when there was perfect separation).

This occurred for nickel and cobalt at an A/O ratio of 1.5 and a H2SO4 concentration of 0.025 M. Even though no cobalt stripping occurred at these conditions only 56.9% of the nickel was stripped which is not considerably high. The second highest value of nickel relative to cobalt stripped (S(%)Ni/S(%)Co) is

332 at an A/O ratio of 0.5 and a H2SO4 concentration of 0.075 M. At these conditions nickel and cobalt are 66.52% and 0.2% stripped, respectively. The best operating conditions for selective stripping of

Stellenbosch University http://scholar.sun.ac.za Batch tests

nickel and cobalt are at an A/O ratio of 0.75 and a H2SO4 concentration of 0.1 M generating a

S(%)Ni/S(%)Co value of 78 with 93.25% and 1.2% of nickel and cobalt being stripped respectively. A major drawback from the stripping tests was the fact that various phase separation issues were encountered as indicated with the shaded area in Table 3-10. A general trend is observed that at low acid concentrations and A/O ratios phase separation problems occur (i.e., when there is not sufficient H+ present in the aqueous phase). This was also the case for the factorial design at both the high and low temperatures. In Table 3-10 all the high values for S(%)Ni/S(%)Co are within the shaded areas. Therefore even though nickel and cobalt may selectively be removed from the organic phase this operation will be limited due to phase separation problems. No literature for the selective stripping of cobalt and nickel from Cyanex 272 was discovered. The phase separation dilemma where a single emulsion forms is a possible explanation.

Referring to Figure 3-19(c) no iron stripping is observed for H2SO4 concentrations equal or less than 0.1

M irrespective of the A/O ratio. This is reflected with all the inf values for S(%)Ni/S(%)Fe and S(%)Co/S(%)Fe indicated in Table 3-10 for these low acid concentrations since iron stripping is zero. The highest stripping for cobalt and nickel at these inf values are at A/O ratios between 2.25-3 and a H2SO4 concentration of 0.1 M where basically all of the nickel and cobalt are removed leaving an organic phase primarily consisting of iron (an A/O ratio of 2.25 is selected since less acid will then be used and a more concentrated nickel and cobalt solution will be generated than at an A/O ratio of 3). This compares well with literature, as shown in Table 2-3. No phase separation problems are encountered at these conditions making this a much more practical operation. Referring to Figure 3-19(d) the pH at which this stripping unit will have to be controlled at is 1.82.

According to Figure 3-8 iron is only completely extracted with Ni-Cyanex 272 at a pH of 3.5 which is higher than 1.82. Therefore the metal content in each respective phase differs as a function of pH during stripping and extraction at equilibrium which means that the metal ions are relatively stable once they are in the organic phase as reflected by the low pH values needed for stripping (the same is true for nickel and cobalt).

From Figure 3-19(c) a H2SO4 concentration of 1 M at an A/O ratio of 0.75 are sufficient to regenerate the organic phase from iron. This is lower than the 1.5 M H2SO4 required for the stripping of iron from Cyanex 302 found in literature (refer to Table 2-4) which is expected since Cyanex 302 is a stronger acid than Cyanex 272. A relatively low A/O ratio is chosen to generate a more concentrated iron solution rather than a dilute solution. The corresponding pH measurement was 1.41.

Stellenbosch University http://scholar.sun.ac.za Modelling and Simulation

CHAPTER 4

4 MODELLING AND SIMULATION

4.1 INTRODUCTION

The data generated from the batch tests with respect to variation in O/A ratio and NaOH addition for the pre-loading section, O/A ratio and nickel loading for the extraction section and A/O ratio and H2SO4 concentration for the stripping section were used to model and develop a process simulation sheet in Microsoft Office Excel® that can be used to approximate operating design variables at equilibrium.

4.2 METHOD

For the pre-loading section the nickel loading model was derived from the equilibrium constant obtained from reaction 13.

[23]

By taking the logarithm on each side and after some mathematical manipulation with respect to the fact that pH is defined as –log [H+] and the distribution coefficient which is defined via equation 1, equation 23 can be rewritten as shown in equation 24:

[24]

where: DNi = Distribution coefficients K = Equilibrium constant

[(RH)2] = Concentration of the extractant/Cyanex 272 (dimer form)

Equation 25 displays the distribution coefficient as a function of percentage extraction (E) and O/A ratio (Swain et al., 2008):

[25]

Substituting equation 25 in equation 24 results in equation 26 after some rearrangement:

Stellenbosch University http://scholar.sun.ac.za Modelling and Simulation

[26]

3 Since the values typically calculated for [(RH)2] are two to three orders of magnitude smaller than the 3 values calculated for the pH term, it was assumed that this term [(RH)2] would have a negligible effect in order to simplify the derivation of the model. Extraction can also be defined as the nickel concentration in the organic phase at equilibrium divided by the initial nickel aqueous concentration multiplied by the O/A ratio and a 100. Equation 26 then simplifies to equation 27 which is rearranged to give equation 28:

[27]

[28]

where: Niorg = Nickel concentration in the organic phase at equilibrium/nickel loading (g/l)

Nioaq = Initial nickel aqueous concentration (g/l)

Assuming that pH is proportional to the relative moles of NaOH added with respect to the initial moles of nickel in the aqueous phase (NaOH/Nio), equation 29 was derived to predict nickel loading.

[29]

A similar equation to equation 29 could not be derived as a function of the operating variables for the loading pH, extraction and stripping data. The sigmoid function, which was also used by Swain et al. (2008) to model cobalt and lithium extraction with Cyanex 272, (see equation 30 and Figure 4-1) was used as origin to model the remaining data. Equation 29 gave a better fit for the nickel loading data than the sigmoid function (R2 value, as calculated with equation 34 in section 4.3, of 0.97 and 0.95 respectively). The sigmoid function also poorly extrapolated the nickel loading data compared to equation 29.

[30]

f(x)

-4 -20 x

Figure 4-1: Characteristic S shape of the sigmoid function

Stellenbosch University http://scholar.sun.ac.za Modelling and Simulation

Referring to Figure 4-1, the sigmoid function has a characteristic S shape which is typical for the pre- loading, extraction and stripping curves that were generated in the previous chapter. Consequently it gave a good fit for the loading pH, extraction pH, cobalt and iron extraction and nickel, cobalt and iron stripping data (Refer to Figures 3-16 [a, c and e] and 3-19 [a, b and c]. The loading pH data are displayed shortly).

Referring to equation 30, it is noticed that the sigmoid function is only two dimensional (in other words pre-loading, extraction and stripping are only modelled as a function of NaOH/Nio, O/A ratio and H2SO4 concentration respectively). After inspection it was observed that the constant, P, varies in a polynomial manner as a function of O/A ratio, nickel loading and A/O ratio for pre-loading, extraction and stripping respectively. Consequently a third dimension was added to equation 30 by expanding the constant P as shown in equation 31 (third and lower order polynomials were found adequate).

[31]

where: f(y,x) = Loading pH, extraction pH, cobalt and iron extraction (%) and nickel, cobalt and iron stripping (%)

K, C,b1-6 = Modelling constants y = O/A ratio (pre-loading), nickel loading (extraction) and A/O ratio (stripping)

x = NaOH/Nio (pre-loading), O/A ratio (extraction) and H2SO4 concentration (stripping)

2 The supplementary expansion (addition of b4 + b5y + b6y to x) was to allow the graph in Figure 4-1 to shift to the left and right to give a better fit where necessary. Most frequently not all the polynomial terms were necessary and equation 31 was simplified by lowering the polynomial order (see Table A-22 in Appendix A which contains all the solved values for K, C and b1-b6 for the different models). Allocating y to be nickel loading and x to be O/A and not vice versa for the extraction section was chosen through a process of trial and error to see which combination results in the best fit and was accordingly chosen as allocated in the above glossary. The same is applicable for the loading pH and stripping models.

For the purpose of developing a process simulation sheet it is important that the models also predict the aqueous concentrations in the exit streams for the extraction and stripping sections. The aqueous product concentration of the pre-loading section can be determined via a mass-balance. Equations 32 and 33 below were used to convert f(y,x)/percentage extracted and stripped into aqueous concentration at equilibrium for the extraction and stripping sections respectively.

(Extraction)

[32]

Stellenbosch University http://scholar.sun.ac.za Modelling and Simulation

(Stripping)

[33]

where: [M] = Metal concentration in the aqueous phase at equilibrium (g/l)

[M]o = Initial metal concentration in the aqueous phase (g/l)

[M]org = Metal loading on the organic phase (g/l)

4.3 RESULTS

The modelling results with respect to nickel loading (equation 29), metal concentration in the aqueous phase for the extraction and stripping sections (equation 32 and 33) and the loading and extraction pH (equation 31) will be considered. Three-dimensional (3D) graphs of each model can be found in Appendix A as well as two-dimensional (2D) graphs which are the same as those that will be consider next but only on separate axes since the curves are frequently close to each other and plotting them on separate axes results in a better visual indication of each fit.

The quality of each model fit was quantified by calculating the R2 value (see equation 34) for the overall 3D surface as well as for various 2D curves at specific O/A ratios, nickel loadings and A/O ratios for the pre-loading, extraction and stripping sections respectively. A R2 value of 1 signifies a perfect model fit, in other words the closer the R2 value is to 1, the better the fit (generally a R2 value higher than 0.95 is regarded as a good fit). The formula used to calculate R2 is shown in equation 34.

[34]

where: SSerr = Sum of squares of residuals

SStot = Total sum of squares

yi = Experimental value y = Average of the experimental values

fi = Modelled value

The predicted/modelled values were also plotted against the experimental values which give a visual indication of how good each fit is. For a perfect fit all of the points should be on a straight line with a y- intercept of 0 and a slope of 1.

Stellenbosch University http://scholar.sun.ac.za Modelling and Simulation

4.3.1 PRE-LOADING

Figure 4-2 shows the pH and nickel loading modelling results for the pre-loading section. The R2 values for each model are found in Table 4-1.

7.7 7.7 7.5 7.5

7.3 7.3 7.1 7.1 6.9

6.9 6.7 6.7 pH 6.5 6.5 6.3 6.3 Predicted pH Predicted 6.1 6.1 5.9 5.9 5.7 5.7 0.8 1.3 1.8 2.3 2.8 3.3 3.8 4.3 5.7 5.9 6.1 6.3 6.5 6.7 6.9 7.1 7.3 7.5 7.7

NaOH/Nio (mol/mol) Experimental pH (a) (b)

6.5 6.5

6.0 6.0 5.5 5.5 5.0 5.0 4.5 4.5 4.0 4.0 3.5 3.5 Nickel loading loading (g/l) Nickel 3.0 3.0 2.5 2.5 Predicted nickel loading (g/l) nickel loading Predicted 0.8 1.3 1.8 2.3 2.8 3.3 3.8 4.3 2.5 3.5 4.5 5.5 6.5 7.5 NaOH/Nio (mol/mol) Experimental nickel loading (g/l) (c) (d)

6.06.5 O/A = 26.67 (experimental) O/A = 26.67 (model) 5.05.5 4.5 O/A = 20 (experimental) O/A = 20 (model) 3.54.0

3.0(g/l)

Nickel Nickel 2.5

loading loading 2.0 O/A = 16 (experimental) O/A = 16 (model) 0.8 1.3 1.8 2.3 2.8 3.3 3.8 4.3 O/A = 13.3 (experimental)NaOH/Nio (mol/mol) O/A = 13.3 (model)

Figure 4-2: Model results for nickel loading and equilibrium pH for the pre-loading section

Table 4-1: R2 values of the models developed for equilibrium pH and nickel loading for the pre-loading section

O/A (ml/ml) 26.67 20.00 16.00 13.33 Overall for 3D model pH 0.97 0.98 0.95 0.95 0.97 Nickel loading 0.70 0.83 0.96 0.97 0.97

Referring to Figure 4-2 (a) it is noticed that the loading pH is predominantly influenced by NaOH/Nio and not by the O/A ratio, resulting in the lines being so close to each other. The model accurately predicts the loading pH behaviour with R2 values higher than 0.95 at all the O/A ratios as well as for the overall 3D model. This is also reflected in Figure 4-2 (b) with all the points being closely positioned on the straight line. Nickel loading is accurately predicted by the model for O/A ratios of 16 and 13.33 with R2

Stellenbosch University http://scholar.sun.ac.za Modelling and Simulation

values of 0.96 and 0.97, respectively. The low R2 value of 0.7 at an O/A ratio of 26.67 is caused by the fact that the nickel loading do not vary significantly at this high O/A ratio and the slightest under or over prediction will have a significant influence on the R2 value which is the case at NaOH/Nio values less than 1.8 mol/mol. The model also over predicts the nickel loading at an O/A ratio of 20 for NaOH/Nio additions less than 1.8 mol/mol by approximately 0.4 g/l and as a result has a R2 value of 0.83. From Figure 4-2 (d) it is clear that a more accurate model prediction is attained for nickel loadings higher than 3.5 g/l with all the points being closer to the straight line than the points at loadings lower than 3.5 g/l. Even so, the overall R2 value of 0.97 signifies a generally good fit, implying that the nickel loading model can be used to accurately predict loadings.

4.3.2 EXTRACTION

The model results for extraction pH and cobalt and iron concentrations in the aqueous phase at equilibrium are considered next. Refer to Figure A-5 in Appendix A where graphs might be unclear.

The R2 values for each fit are displayed in Table 4-2.

6.0 6.0

5.5 5.5

5.0 5.0

4.5 4.5 pH

4.0 4.0 Predicted pH Predicted

3.5 3.5

3.0 3.0 0.25 0.75 1.25 1.75 2.25 2.75 3.25 3.0 3.5 4.0 4.5 5.0 5.5 6.0

O/A Experimental pH (a) (b)

1.2 1.2

1.0 1.0

0.8

0.6 0.6 Co (g/l) Co 0.4 0.4 Predicted Co (g/l) Predicted 0.2 0.2

0.0 0.0 0.25 0.75 1.25 1.75 2.25 2.75 3.25 0.0 0.2 0.4 0.6 0.8 1.0 1.2

O/A Experimental Co(g/l) (c) (d)

Stellenbosch University http://scholar.sun.ac.za Modelling and Simulation

3.0 3.0

2.5 2.5

2.0 2.0

1.5 1.5 Fe (g/l) Fe 1.0 1.0 Fe (g/l) Predicted

0.5 0.5

0.0 0.0 0.25 0.75 1.25 1.75 2.25 2.75 3.25 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Experimental Fe (g/l) (f) O/A (e) Ni loading = 0.56g/l (experimental) Ni loading = 0.56g/l (modell) Ni loading = 1.97g/l (experimental) Ni loading = 1.97g/l (model) Ni loading = 2.94g/l (experimental) Ni loading =2.94g/l (modell) pH 3.0 Ni loading = 3.95g/l (experimental) Ni loading = 3.95g/l (model) 0.25 0.75 1.25 1.75 2.25 2.75 3.25 Ni loading = 4.69g/l (experimental) O/A Ni loading = 4.69g/l (model) Ni loading = 5.73g/l (experimental) Ni loading = 5.73g/l (model)

Figure 4-3: Model results for equilibrium pH and cobalt and iron concentration in the pregnant nickel solution at equilibrium for the extraction section

Table 4-2: R2 values of the models developed for equilibrium pH and cobalt and iron concentration in the aqueous phase at equilibrium

Nickel loading (g/l) 0.56 1.97 2.94 3.95 4.69 5.73 Overall for 3D model pH 0.06 0.90 0.89 0.89 0.96 0.89 0.94 Co -6.23 0.99 1.00 0.99 0.99 1.00 0.99 Fe 0.98 1.00 0.99 0.99 0.99 0.99 0.99

From Figure 4-3 (b) it is clear that the model gives a better fit for pH values higher than 4 with the points being closer to the straight line than for pH predictions lower than 4. Referring to Figure A-5 it is visually observed that the pH-model gives a better fit as the nickel loading increases as a function of O/A ratio. The R2 value of 0.06 at a loading of 0.56 g/l is once again caused by the fact that the pH data at this loading remains steady and a slight deviation in the model prediction will have severe influence on the R2 value. The same applies to the R2 value of -6.23 for cobalt at a nickel loading of 0.56 g/l.

The models perform exceptionally well when predicting the aqueous concentration of cobalt and iron at equilibrium with R2 values higher and equal to 0.98 for all scenarios apart from the single exception regarding cobalt just mentioned. This exception is also responsible for all the points that deviate from the straight line for cobalt concentrations higher than 1 g/l in Figure 4-3 (d). Referring to Figure 4-3 (c and e) it is noted that the aqueous cobalt concentration at equilibrium is more sensitive for variations in O/A ratio and nickel loading than the aqueous iron concentration at equilibrium, in other words there is a sudden drop in cobalt concentration whereas the iron concentration has a more gradual decrease as the O/A ratio and nickel loading increases. This sudden drop is accurately predicted by the model as indicated in Figure A-5 and can therefore be used to accurately predict the cobalt and iron behaviour in the extraction section.

Stellenbosch University http://scholar.sun.ac.za Modelling and Simulation

4.3.3 STRIPPING

The modelling results for the nickel, cobalt and iron concentration in the aqueous phase at equilibrium are shown in Figure 4-4 (refer to Figure A-6 where graphs may appear unclear). The R2 values of each model are found in Table 4-3.

5.5 5.5

4.5 4.5

3.5 3.5

Ni (g/l) Ni 2.5 2.5

1.5 Ni (g/l) Predicted 1.5

0.5 0.5 0.00 0.05 0.10 0.15 0.20 0.25 0.5 2.5 4.5 H₂SO₄ (M) Experimental Ni (g/l) (a) (b)

1.4

1.4 1.2 1.2 1.0 1.0

0.8 0.8 0.6 0.6 Co (g/l) Co 0.4 0.4 Predicted Co (g/l) Predicted 0.2 0.2 0.0 0.0 0.00 0.05 0.10 0.15 0.20 0.25 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

H₂SO₄ (M) Experimental Co (g/l) (c) (d) 4.5 4.5

4.0 4.0 3.5 3.5 3.0 3.0 2.5 2.5 2.0 2.0 Fe (g/l) (g/l) Fe 1.5 1.5 1.0 1.0 Predicted Fe (g/l) 0.5 0.5 0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

H₂SO₄ (M) Experimental Fe (g/l) (e) (f) A/O = 0.5 (experimental) A/O = 0.5 (model) 5

A/O = 0.75 (experimental) A/O = 0.75 (model)

pH A/O = 1 (experimental) A/O = 1 (model) A/O = 1.5 (experimental) A/O = 1.5 (model) 0 0.00 A/O0.10 = 2.250.20 (experimental) 0.30 0.40 0.50 0.60 A/O0.70 = 2.250.80 (model) 0.90 1.00 A/O = 3 (experimental) H₂SO₄ (M) A/O = 3 (model)

Figure 4-4: Model results for nickel, cobalt and iron concentration in the aqueous phase at equilibrium for the stripping section

Stellenbosch University http://scholar.sun.ac.za Modelling and Simulation

Table 4-3: R2 values of the models developed for nickel, cobalt and iron concentration in the aqueous phase at equilibrium for the stripping section

A/O (ml/ml) 0.50 0.75 1.00 1.50 2.25 3.00 Overall for 3D model Ni 0.95 0.93 0.95 0.72 0.68 0.24 0.99 Co 0.98 0.98 0.98 0.93 0.89 0.78 0.98 Fe 0.88 0.98 0.96 0.96 0.95 0.90 0.94

The nickel model accurately fits the experimental data with an overall R2 value of 0.99. The relatively low R2 values of 0.72, 0.68 and 0.24 at A/O ratios of 1.5, 2.25 and 3, respectively, are once again caused by the relatively constant experimental data as discussed previously. Referring to Figure 4-4 (a) it is clear that the model accurately predicts these trends as well.

Referring to Figure 4-4 (d) the cobalt model is also performing superbly with the majority of points being close to the straight line. This is also reflected in the high R2 value of 0.98 for the model.

The iron model is performing the worst of the three models when comparing Figure 4-4 (f) to Figures 4-4 (b and d). This is also reflected in the R2 value of 0.94 compared to 0.98 and 0.99 for cobalt and nickel, respectively. It is noted that the iron model over predicts the iron concentration at acid concentrations less than 0.2 M and larger than 0.5 M for an A/O ratio of 0.5. This is reflected in Figure 4-4 (f) for experimental iron concentrations equal to 0 g/l and larger than 3 g/l. Besides this, the model also accurately predicts iron concentrations at the remaining A/O ratios, with R2 values higher and equal to 0.9.

4.4 SIMULATION

4.4.1 SIMULATION SHEET

Figure 4-5 shows the simulation sheet that was developed in Microsoft Excel from the derived models. The dark and light grey boxes represent the organic and aqueous phases respectively. The italic values are the input values and the remaining values are the output values. The input values include the metal concentrations and flow rates of the pregnant solution, the O/A ratio for the pre-loading and extraction sections, A/O ratio for the stripping sections, the relative moles of NaOH being added with respect to nickel in the aqueous feed (NaOH/Nio), the NaOH concentration used for the pre-loading section and the H2SO4 concentration of the two streams being fed to the stripping sections. The output values include the flow rate of all the remaining feed and product streams, percentage extracted and stripped at the relevant stages, the pH at the pre-loading and extraction stages and the metal concentration in the remaining streams.

The stream concentrations in the flow sheet that are not determined by the models are resolved via mass balance calculations. These include the nickel concentration of the pre-loading aqueous product solution as well as the metal loading of the organic phase after extraction and stripping. No models were developed for the nickel behaviour in the pregnant solution at the extraction stage or for the sodium behaviour in pre-loading and extraction aqueous product solutions. For the sake of simplicity the

Stellenbosch University http://scholar.sun.ac.za Modelling and Simulation

amount of nickel extracted from the pregnant solution can be selected as indicated by the italic value in the extraction box/unit. It is also assumed in the sheet that no sodium ends up in the pregnant solution and that all the sodium entering the pre-loading stage is exiting at the aqueous product of this stage. When not all the metal is stripped at the final stage it will be indicated in the organic feed to the pre- loading stage.

0.07 l/min 3.38 l/min 1.13 l/min H SO H SO NaOH (g/l) 120 2 4 0.10 2 4 1.00 (M) (M) NaOH/Nio 2.50 (mol/mol)

1.50 l/min 1.50 l/min 1.50 l/min 1.50 l/min Ni (g/l) 0.00 Ni (g/l) 3.18 Ni (g/l) 2.86 Ni (g/l) 0.00 Pre-loading Extraction Stripping Stripping Co (g/l) 0.00 Co (g/l) - Co (g/l) 0.63 Co (g/l) 0.00 Fe (g/l) 0.01 Fe (g/l) - Fe (g/l) 1.99 Fe (g/l) 1.99 Na (g/l) - pH 6.98 Na (g/l) - pH 4.58 Na (g/l) - Na (g/l) - O/A 25.00 O/A 1.50 A/O 2.25 A/O 0.75 0.06 l/min

Ni (g/l) 80 ENi (%) 98.8 ENi (%) -0.6 SNi (%) 100.0 SNi (%) 100.0

Co (g/l) 0.06 ECo (%) 94.5 SCo (%) 100.0 SCo (%) 100.0

Fe (g/l) 0.01 EFe (%) 99.7 SFe (%) 0.3 SFe (%) 99.7 Na (g/l) -

1.00 l/min 1.13 l/min Ni (g/l) 80 Ni (g/l) 0.00 Co (g/l) 1 Co (g/l) 0.00 Fe (g/l) 3 Fe (g/l) 2.64 Na (g/l) - Na (g/l) -

3.38 l/min Ni (g/l) 1.27 Co (g/l) 0.28 Fe (g/l) 0.00 Na (g/l) -

0.94 l/min Ni (g/l) 80 Co (g/l) 0.06 Fe (g/l) 0.01 Legend Na (g/l) - Input values Output values 0.13 l/min Aqueous phase Ni (g/l) 0.44 Organic phase Co (g/l) - Fe (g/l) - Na (g/l) 36.80

Figure 4-5: Process simulation sheet in Microsoft Excel

The process conditions in Figure 4-5 are set as they were selected from the batch tests (nickel extraction was chosen as -0.6%). This simulation accurately resembles what was observed from the batch tests with respect to cobalt (94.9%) and iron (99.7%) extraction, selective stripping of cobalt (100%) and nickel (100%) from iron (0.3%) at the first stripping stage and iron stripping at the final stage (99.7%).

4.4.2 MODEL LIMITATIONS

The limitations associated with the models developed for the simulation sheet in Figure 4-5 are considered next. In order to do this, model predictions were made outside the original range of the experimental conditions used to generate the equilibrium data.

Stellenbosch University http://scholar.sun.ac.za Modelling and Simulation

Table 4-4, 4-5 and 4-6 reveal the extrapolated model results for the pre-loading, extraction and stripping sections, respectively. Predictions made at the maximum, minimum and selected intermediate conditions are also included in these Tables and are indicated within the shaded areas. The experimental results at the maximum and minimum conditions are also included within brackets.

Table 4-4: Extrapolated model results for the pre-loading section

O/A ratio

7 13.33 16 20 26.67 35

Nickel loading (g/l)/Nickel extraction (%) 0.5 2.41/19.1 2.05/31.0 1.93/35 1.78/40.2 1.57/47.3 1.37/54.1

0.83 4.23/33.5 3.25/48.9 (3.28/49.52) 2.96/53.5 2.61/59 2.18/65.7 1.81/71.6

1 5.39/42.7 3.89/58.6 3.48/63 (3.37/60.97) 3.01/68 2.45/73.9 1.99/78.8

1.24 7.12/56.4 4.72/71.1 4.13/74.7 3.48/78.7 (3.05/68.93) 2.76/83.1 2.19/86.6 NaOH/Nio 1.66 9.61/77.3 5.65/86.6 4.82/88.6 3.94/90.7 3.03/92.8 (2.81/84.65) 2.35/94.4 (mol/mol) 1.99 10.93/87.9 6.09/93.3 (6.5/97.96) 5.13/94.3 4.15/95.4 3.15/96.5 2.42/97.3

2.39 11.97/94.8 6.45/97.2 5.4/97.7 (5.53/100) 4.34/98.1 3.27/98.6 2.5/98.9

2.99 12.46/98.6 6.58/99.3 5.49/99.4 4.4/99.5 (4.42/100) 3.3/99.6 2.52/99.7

3.98 12.61/99.99 6.63/99.99 5.52/99.9 4.42/100 3.31/100 (3.32/100) 2.53/100

5 12.63/100 6.63/100 5.52/100 4.42/100 3.31/100 2.53/100 Loading pH 0.5 6.24 5.39 5.04 4.55 3.78 2.95

0.83 6.44 5.77 (5.76) 5.49 5.08 4.44 3.7

1 6.54 5.95 5.71 (5.78) 5.35 4.77 4.09

1.24 6.67 6.19 5.99 5.7 (5.86) 5.21 4.64 NaOH/Nio 1.66 6.87 6.56 6.43 6.24 5.92 (5.99) 5.53 (mol/mol) 1.99 6.99 6.8 (6.9) 6.72 6.6 6.4 6.15

2.39 7.12 7.03 7 (7.2) 6.95 6.86 6.75

2.99 7.23 7.26 7.27 7.28 (7.4) 7.3 7.33

3.98 7.25 7.29 7.31 7.34 7.38 (7.48) 7.44

5 7.05 6.91 6.85 6.76 6.61 6.42

Referring to Table 4-4, the model realistically extrapolates the nickel loading in that it increases and decreases to sensible maximum and minimum loadings as NaOH/Nio increases and decreases and the O/A ratio decreases and increases, respectively (percentage extraction does not exceed 100% at the highest O/A ratios and NaOH/Nio values). The nickel loading predictions at the maximum and minimum experimental conditions seldom differ more than 0.4 g/l from the experimental results. Consequently the nickel loading model can be used to predict loadings beyond the borders of the original experimental conditions.

Referring to the loading pH in Table 4-4, the model unrealistically predicts the pH at NaOH/Nio values higher than 3.98 since the pH decreases as NaOH/Nio increases. Apart from this, the model provides sensible predictions beyond the original experimental borders in that pH decreases as the O/A ratio increases and NaOH/Nio decreases. The differences between the model predictions and experimental pH do not differ more than 0.2 at the highest and lowest values in the shaded areas. Therefore the pH

Stellenbosch University http://scholar.sun.ac.za Modelling and Simulation

model can be used to forecast a loading pH outside the shaded area except at NaOH/Nio values higher than 3.98.

Table 4-5: Extrapolated model results for the extraction section

O/A ratio

0.15 0.25 3 4

Coaq (g/l)/Cobalt extraction (%)

0.3 1/0.4 1/0.7 1/8.6 1/11.5

0.56 1/0.4 1/0.7 (1.07/4.38) 1/8.8 (1.06/5.19) 0.98/12.8 Ni-loading (g/l) 5.73 1/0.5 1/0.9 (1.09/2.24) 0.03/97.1 (0.03/97.1) 0.03/97.2

6.3 1/0.5 1/0.9 0.03/97.1 0.03/97.2

Feaq (g/l)/Iron extraction (%) 0.3 2.62/12.9 2.59/14.2 1.03/68.7 0.52/84.7

0.56 2.61/13.5 2.56/15.3 (2.64/11.58) 0.43/86.8 (0.53/82.08) 0.13/96.1 Ni-loading (g/l) 5.73 2.05/31.9 1.43/52.8 (1.47/50.57) 0/100 (0/100) 0/100

6.3 1.97/34.6 1.27/57.9 0/100 0/100 Extraction pH

0.3 3.4 3.38 2.84 2.64

0.56 3.21 3.2 (3.16) 3.14 (3.19) 3.12 Ni-loading (g/l) 5.73 3.37 3.63 (3.77) 5.55 (5.58) 5.56

6.3 3.41 3.68 5.56 5.57

From Table 4-5, the model realistically extrapolates cobalt and iron extractions so that both increase and decrease as the nickel loading and O/A ratio increase and decrease, respectively, (iron and cobalt extractions do not exceed 97.1% and 100%, respectively, at the maximum nickel loading and O/A ratio). The cobalt and iron concentrations in the shaded areas rarely differ more that 0.1 g/l. It is therefore concluded that the model can be used to realistically predict cobalt and iron extractions beyond the borders of the shaded areas.

At nickel loadings equal to and less than 0.56 g/l the model does not resemble the expected pH behaviour for the extraction section which normally increases and decreases as the O/A ratio and nickel loading increase and decrease, respectively. Instead an irregular trend is observed where the pH increases from 3.38 to 3.4 and decreases from 2.84 to 2.64 as the O/A ratio decreases from 0.25 to 0.15 and increases from 3 to 4, respectively, at a nickel loading of 0.3 g/l. On the other hand, the pH decreases from 3.21 to 3.12 as the O/A ratio increases from 0.15 to 4 at a nickel loading of 0.56 g/l. This fluctuation, however, remains within realistic boundaries of 2.64 and 3.4 which are typical pH values that can be expected at nickel loadings less than 0.56 g/l. Apart from this, the model rationally extrapolates the extraction pH so that it increases as nickel loading and O/A ratio increase. The experimental and predicted pH also do not differ more than 0.1 in the shaded area. The model can therefore be used to extrapolate realistic pH values for nickel loadings higher than 0.56 g/l.

Stellenbosch University http://scholar.sun.ac.za Modelling and Simulation

Table 4-6: Extrapolated model results for the stripping section

A/O ratio

0.25 0.5 3 4

Niaq (g/l)/Nickel stripped (%) 0.01 1.36/11.4 0.68/11.5 0.12/12.3 0.09/12.6 0.025 1.88/15.9 1.28/21.7 (1.19/20.02) 0.92/92.8 (0.92/93.5) 0.73/98.4

H2SO4 (M) 1 11.84/100 5.92/100 (5.76/99.65) 0.99/100 (0.96/99.96) 0.74/100 4 11.84/100 5.92/100 (5.8/100) 0.99/100 (1/100) 0.74/100 5 11.84/100 5.92/100 0.99/100 0.74/100

Coaq (g/l)/Cobalt stripped (%) 0.01 0/0.1 0/0 0/0 0/0 0.025 0/0 0/0.1 (0.01/0.47) 0.14/55.1 (0.14/55.75) 0.18/96.4

H2SO4 (M) 1 0/0 1.5/100 (1.3/91.05) 0.25/100 (0.27/100)) 0.19/100 4 0/0 1.5/100 (1.45/100) 0.25/100 (0.24/99.98) 0.19/100 5 0/0 0/0 0.25/100 0.19/100

Feaq (g/l)/Iron stripped (%) 0.01 0.66/7.9 0.15/3.6 0/0 0/0 0.025 0.7/8.4 0.17/4.1 (0/0) 0/0 (0/0) 0/0

H2SO4 (M) 1 7.01/84..2 4.12/99.1 (3.9/98.2) 0.69/99.8 (0.69/99.7) 0.52/99.8 4 8.3/99.8 4.15/100 (4.1/100) 0.69/100 (0.6/100) 0.52/99.8 5 8.3/99.8 4.15/99.8 0.69/99.8 0.52/99.8

Referring to Table 4-6, the model realistically extrapolates the nickel concentration in the aqueous phase during stripping since the nickel concentration increases and decreases as the H2SO4 concentration increases and decreases and as the A/O ratio decreases and increases respectively. The modelled and experimental nickel concentrations do not differ more than 0.2 g/l at the maximum, minimum and intermediate experimental H2SO4 concentrations and A/O ratios. Consequently the model can be used to predict realistic values for the quantity of nickel stripped beyond the borders of the shaded area.

At an A/O ratio of 0.5 and H2SO4 concentration of 5 M, the model unrealistically predicts that no cobalt is stripped. Apart from this, the expected trend that percentage cobalt stripped increases and decreases as the acid concentration and A/O ratio increases and decreases is observed outside the shaded area.

Referring to an A/O ratio of 0.5 at the intermediate H2SO4 concentration of 1 M, the model predicts that most of the cobalt is stripped and realistically also predicts this at 4 M; hence the sharp decline only occurs for acid concentrations higher than 4 M. The maximum difference between the experimental and predicted aqueous cobalt concentrations is 0.2 g/l. It is therefore concluded that up to H2SO4 concentrations of 4 M, the model can be used to realistically extrapolate cobalt stripping.

Referring to iron stripping in Table 4-6, the model over- and under- predicts the iron concentration in the aqueous phase at H2SO4 concentrations of 0.025 M and 4 M, respectively, relative to the experimental data with a maximum of 0.17 g/l. The expected trend where the percentage iron stripped increases and decreases as the acid concentration and A/O increase and decrease, respectively, is however predicted by the model outside the shaded area. The model can therefore be use to

Stellenbosch University http://scholar.sun.ac.za Modelling and Simulation

extrapolate iron stripping. It is however important to notice that this will be a slight over- and under- prediction of not more than 0.17 g/l at the higher and lower acid concentrations, respectively.

4.4.3 QUANTIFYING THE NUMBER OF STAGES

The simulation sheet in Figure 4-5 can also be used to determine the number of stages to reach desired nickel-loadings, percentage cobalt and iron extraction and percentage nickel, cobalt and iron stripped at the pre-loading, extraction and stripping sections, respectively. Since the models allow for variation in the inlet aqueous and organic concentrations (see equations 29, 32 and 33), the concentration of the aqueous or organic product streams become the new feed concentration of the next stage and is typed into the simulation sheet accordingly (this is done after determining it from a simulation for the previous stage).

Two different scenarios were considered. During the first simulation, generating a nickel loading of 3.2 g/l at an O/A ratio of 25 and different pH values from an 80 g/l nickel aqueous solution were considered with a counter-current configuration as explained in Figure 2-8. Table 4-7 shows the predicted nickel concentration in the aqueous product at each stage. As mentioned in section 3.3.1.5, even though controlling the nickel loading at various stages is possible from a theoretical point of view, this is not always a practical operation and was consequently not considered with respect to the number of stages at different operating conditions. Another approach of how the simulation sheet can be helpful is to consider A/O ratios and H2SO4 concentrations where complete cobalt and nickel stripping is not attained. This reduces iron stripping as well and the result is a purer cobalt and nickel aqueous product by using multiple stages. As mentioned in section 3.3.2.3, low acid concentrations and A/O ratios where complete cobalt and nickel stripping was not reached, resulted in phase separation problems which limits this approach. The final scenario therefore considered was to use different acid concentrations at the selected A/O ratio of 0.75 to strip a 2 g/l iron-loaded organic solution (see Table 4-8).

Table 4-7: Nickel concentration (g/l) in the aqueous product at different stages to generate a nickel loading of 3.2 g/l from an 80 g/l nickel feed solution at an O/A ratio of 25 as predicted by the simulation sheet pH 1st stage 2nd stage 3rd stage 4th stage 5th stage 5.5 7.295 1.014 0.148 0.022 0.003 6 3.467 0.255 0.019 0.001 0.000 6.5 1.411 0.047 0.002 0.000 0.000 7 0.409 0.004 0.000 0.000 0.000

Table 4-8: Iron concentration (g/l) in the organic phase at different stages as predicted by the simulation sheet to strip a 2 g/l iron loaded organic phase at an A/O ratio of 0.75

st nd rd th th H2SO4 (M) 1 stage 2 stage 3 stage 4 stage 5 stage 0.3 1.243 0.772 0.480 0.298 0.185 0.4 0.639 0.204 0.065 0.021 0.007 0.5 0.239 0.029 0.003 0.000 0.000 0.75 0.016 0.000 0.000 0.000 0.000 1 0.005 0.000 0.000 0.000 0.000

Referring to Table 4-7, it is clear that the nickel concentration in the aqueous product decreases at each stage as the pH increases. 0.022 g/l of nickel for example is left in the aqueous product solution at the

Stellenbosch University http://scholar.sun.ac.za Modelling and Simulation

4th stage for a pH of 5.5 compared to 0.001 g/l at the same stage but when controlling the pH at 6. From Table 4-8, the same goes for the iron concentration in the organic product which decreases at each stage as the H2SO4 concentration increases.

During stripping a fresh H2SO4 solution is fed to each stage compared to the pH being controlled during pre-loading. Consequently there will be various aqueous products compared to the single aqueous raffinate stream for pre-loading. The single stage at a pH of 7 and 1 M H2SO4 solution for the nickel pre- loading and iron stripping sections are, however, still selected as operating conditions (the iron concentration in the 2nd stage will be lower since the model under-predicts iron concentration as discussed in section 4.4.2). This is justified by the fact that the quantity of NaOH required to extract nickel in several stages will be similar to the quantity of NaOH in a single stage. This reasoning is also applicable to the quantity of H2SO4 that will be required to strip iron and that adding additional stages will not add any value with respect to the purity of the product as it would have done for cobalt and nickel stripping as discussed previously.

4.4.4 SENSITIVITY ANALYSIS

The sensitivity with respect to the extent to which cobalt and iron are extracted and nickel, cobalt and iron are stripped for variations in the feed concentrations at the selected operating conditions is considered next. Prior to varying the inlet concentrations in the simulation sheet, different McCabe- Thiele diagrams will be discussed to obtain a better understanding of the simulated results to follow.

Figure 4-6 reveals the McCabe-Thiele diagrams for iron extraction (feed concentrations of 2 g/l, 4 g/l and 5.5 g/l) at O/A feed ratios of 1.5 (a) and 2 (b) with nickel loadings of 2.94 g/l and 3.95 g/l.

5 5

4 4

3 3 organic (g/l) organic -

organic (g/l) organic 2 2 - Fe Fe

1 1

0 0 0 1 2 3 4 5 6 0 1 2 3 4 5 6 Fe-aqueous (g/l) Fe-aqueous (g/l) (a) (b)

5.0

- 0.0

… Ni-organic = 2.94g/l Ni-organic = 3.95g/l or g Fe 0.0 0.5 1.0 1.5 2.0Fe -aqueous2.5 3.0 3.5 (g/l) 4.0 … 4.5 5.0 5.5 6.0

Figure 4-6: McCabe-Thiele diagrams for nickel loadings of 2.94 g/l and 3.95 g/l with an O/A feed ratio of 1.5 (a) and 2 (b) to extract 4 g/l and 5.5 g/l of iron

Stellenbosch University http://scholar.sun.ac.za Modelling and Simulation

Referring to Figure 4-6 (a), decreasing and increasing the iron concentration in the feed stream from the original 3 g/l to 2 g/l and 4 g/l, respectively, does not significantly influence iron extraction (concentration in the product stream in both cases is close to 0 g/l) at a nickel loading of 2.94 g/l. Consequently it will also not significantly influence iron extraction at the selected nickel loading of 3.2 g/l. Increasing the feed concentration to 5.5 g/l, however, requires an additional extraction stage at a nickel loading of 2.94 g/l and O/A ratio of 1.5. This additional stage can be prevented by either increasing the nickel loading to 3.95 g/l as shown in Figure 4-6 (a), or by increasing the O/A ratio to 2 as shown in Figure 4-6 (b). The McCabe-Thiele diagram for cobalt extraction also has a very steep slope at these nickel loadings (refer to Figure 3-16 (b). Consequently cobalt and iron extraction will not be sensitive for increases in the aqueous feed concentrations. Upper limits, however, do exist like the 5.5 g/l for iron extraction at an O/A ratio of 1.5 and nickel loading of 2.94 g/l (this will be higher for the selected nickel loading of 3.2 g/l though). The upper limit determined for cobalt in the same manner as shown in Figure 4-6 for iron at a nickel loading and O/A ratio of 2.94 g/l and 1.5 is approximately 1.5 g/l. This will also be higher at the selected nickel loading of 3.2 g/l.

The McCabe-Thiele diagrams for the nickel, cobalt and iron stripping also have very steep slopes. Figure 4-7, for example, shows the McCabe-Thiele diagram for nickel at the selected A/O ratio and acid concentration of 2.25 and 0.1 M for the first stripping section.

Ni aq (g/l) aq Ni 2

0 0 1 2 3 4 5 6 Ni org (g/l)

Figure 4-7: McCabe-Thiele diagram for stripping 2 g/l and 5 g/l of nickel with 0.1 M H2SO4 at an A/O feed ratio of 2.25

Referring to Figure 4-7, decreasing and increasing the nickel concentration to 2 g/l and 5 g/l, respectively, compared to the originally selected 2.69 g/l (see section 3.3.2.3), does not significantly influence the nickel concentration in the product stream (close to 0 g/l in both cases). Likewise it can be proven that increasing the cobalt and iron organic loading to 5 g/l at the selected acid concentrations and A/O ratios will not significantly influence the loading of the organic product. Increasing all three nickel, cobalt and iron organic loadings beyond 5 g/l will result in third phase formation caused by the ligand being overloaded. Realistic upper limits therefore do not exist as for cobalt and iron extraction.

The sensitivity analysis in Table 4-9, that was conducted with the simulation sheet, reveals the sensitivity with respect to extraction and stripping for variations in the feed concentration.

Stellenbosch University http://scholar.sun.ac.za Modelling and Simulation

Table 4-9: Sensitivity analysis with respect to variations in the feed concentration to the extraction and stripping sections as predicted by the simulation sheet

Cobalt and iron extraction at an O/A ratio of 1.5 and nickel loading of 3.2 g/l Cobalt in aqueous Cobalt in aqueous Iron in aqueous Iron in aqueous Run feed (g/l) product (g/l) feed (g/l) product (g/l) 1 1 0.057 3 0.008 2 1.5 0.085 4 0.01 3 2 0.113 5 0.013 Nickel and cobalt stripping at an A/O ratio of 2.25 with 0.1 M H2SO4 Nickel in organic Nickel in organic Cobalt in organic Cobalt in organic Iron in organic Iron in organic feed (g/l) product (g/l) feed (g/l) product (g/l) feed (g/l) product (g/l) 4 2.96 0.0006 0.67 0.0003 2 1.9939 5 4 0.0009 1 0.0005 4 3.9877 6 5 0.0011 3 0.0014 5 4.9847

Iron stripping at an A/O ratio of 0.75 with 1 M H2SO4 Iron in organic Iron in organic feed (g/l) product (g/l) 7 2 0.005 8 4 0.011 9 5 0.013

Referring to Table 4-9, it is observed that increasing the feed concentration does, in fact, increase the metal content in the product streams. This is expected by looking at Figures 4-6 and 4-7 which reveal that even though the extraction isotherms generally have very steep slopes, they are in fact not vertical and that an increase in the feed concentration will increase the product concentration. If the absolute concentrations in the respective aqueous and organic product streams are of primary importance, then apart from cobalt yielding an aqueous product of 0.113 g/l at a feed concentration of 2 g/l compared to 0.057 g/l and 0.085 g/l at feed concentrations of 1 g/l and 1.5 g/l, respectively; the simulation sheet predicts that the selected operating variables for the extraction circuit are robust towards variations in the feed concentration for the extraction and stripping sections.

Stellenbosch University http://scholar.sun.ac.za Continuous tests

CHAPTER 5

5 CONTINUOUS TESTS

5.1 INTRODUCTION

A wide range of multistage extractor/contactor designs exist in solvent extraction. Essentially all of them can be categorized into differential or stage-wise contactors (Habashi, 1999). With differential contactors the organic and aqueous phases are continuously in contact with each other and complete phase separation is only reached at the end of the unit. Typical examples include spray, packed, perforated plate, pulse and rotating disk columns (Perry and Green, 2007; Richardson et al., 2002; Habashi, 1999). The height of these columns depends on the number of stages required. Stage-wise contactors consist of a series of physical stages in which the phases are mixed and separated at each stage. The most common example is the mixer-settler system which consists of a series of mixing and settling chambers. The organic and aqueous phases are mixed with a rotating impeller in the mixing chamber and then flow to the settling chamber where phase separation occurs.

From the batch tests in chapter 3 it was observed that the extraction and stripping kinetics are fast with equilibrium being reached within 5 and 10 minutes, respectively. It was also noticed that a single stage is sufficient to attain the desired metal extraction and stripping at the chosen operating conditions. Consequently a mixer-settler system was selected instead of a column design for the continuous tests since column designs are frequently used when several stages are required to extract or strip a single metal and the mixer-settler system will not be too large based on the fast kinetics which can be problematic when floor space is limited.

Four continuous tests were conducted to examine the proposed process conditions from the simulation sheet. In the first two tests pre-loading and extraction were considered and in the final two tests stripping was considered. The sole purpose of the continuous tests was to evaluate the batch and simulated results. Pilot and plant scale tests will differ from the tests conducted in this section.

5.2 MIXER-SETTLER SETUP

A four-stage mixer-settler setup having 320 ml capacity in the mixers and 1350 ml capacity in the settlers, manufactured from borosilicate/Pyrex glass, was purchased from SX Kinetics Canada. Clear and yellow tubing connecting each stage were used for the aqueous and organic phases respectively. The yellow organic tubing is specifically designed for organic solutions (Tygon B-44-3) and the clear tubing is made from Tygon F-4040-A suitable for aqueous solutions. The connectors between the tubing and mixing and settling chambers were manufactured from polypropylene. Peristaltic pumps with a capacity of 20-200 ml/min were used to feed the organic and aqueous phases to the system. Four variable-speed

Stellenbosch University http://scholar.sun.ac.za Continuous tests

mixer agitators (40-2000 rpm) with curved-blade impellers manufactured from polypropylene were used for agitation in the mixers (a stirring speed of 1000 rpm was used for all the tests). Seven litre stainless steel containers were used as organic and aqueous inventories and were pre-heated by placing them in a heating bath to control the temperature. Temperature and pH monitors were mounted above each stage and were used to monitor the temperature and pH respectively. In order to control the pH for the pre-loading stage, a flow meter was connected to a tube attached to a valve at the bottom of a chamber filled with NaOH solution, the flow of which was adjusted depending on the flow meter reading. Gravity was used to ensure flow by mounting the NaOH chamber approximately 20cm above the mixing chamber.

The O/A ratio at each stage during operation was determined by sampling a 10 ml aliquot with a pipette from the mixing chamber into a measuring cylinder which revealed the O/A ratio once the two phases separated. Each pump’s flow rate was calibrated by measuring the time it took to fill a 200 ml flask at various capacities (0-100%) as indicated on the pump knob. These flow rates were confirmed during operation by measuring the time it took to fill a 200 ml flask from each product stream.

Figure 5-1 illustrates the flow through the mixers and settlers.

Mixer

Settler Aqueous overflow weir (height controls organic aqueous interface in settler)

Organic Organic inlet from Aqueous inlet from advanced outlet previous stage or previous stage or organic inventory aqueous inventory

Aqueous advanced outlet

Figure 5-1: Schematic illustration of the organic and aqueous flow through the mixers and settlers

Referring to Figure 5-1 the organic and aqueous phases are fed to the bottom of each mixer where after the dispersion advances through the mixer overflow weir to the settling chamber where the aqueous and organic phases separate and exit at the bottom and top, respectively. The position of the organic- aqueous interface in the settlers is controlled by the position of the aqueous over-flow weir which can be raised or lowered.

See Figures A-7 and A-8 in Appendix A for photographs of the setup.

Stellenbosch University http://scholar.sun.ac.za Continuous tests

5.3 MIXER-SETTLER CONFIGURATIONS

Figures 5-2 to 5-5 show the four configurations and process conditions that were used to test the predictions made by the simulation sheet. The dotted lines signify the organic streams and the solid lines the aqueous streams.

nd Pre-loading (1st stage) Extraction (2 stage) O/A = 1 O/A = 1.5 Residence time in mixer = 2.3 min Residence time in mixer = 2.7min Residence time in settler = 9.6 min Residence time in settler = 11.5min

15 ml/min 40g/l NaOH

Settler Settler 1 2 Mixer Mixer 1 2

70 ml/min

55 ml/min 10.74g/l Ni

47 ml/min 80g/l Ni 1.05g/l Co 3.6g/l Fe

nd Pre-loading (1st stage) Extraction (2 stage) O/A = 0.67 O/A = 1.5 Residence time in mixer = 2.01 min Residence time in mixer = 2.99 min Residence time in settler = 8.5 min Residence time in settler = 12.6 min

18 ml/min 40g/l NaOH

Settler Settler 1 2 Mixer Mixer 1 2

64 ml/min

77 ml/min 15g/l Ni

43 ml/min 82g/l Ni 1.02g/l Co 3.07g/l Fe

Stellenbosch University http://scholar.sun.ac.za Continuous tests

Ni and Co stripping (1st stage) Fe stripping (2nd stage) A/O = 2.3 A/O = 0.72 Residence time in mixers = 3.8min Residence time in mixer = 3.7min Residence time in settlers = 16.2min Residence time in settler = 15.7min

Settler Settler Settler 1 2 3 Mixer Mixer Mixer 50 ml/min 1 2 3 0.94g/l Ni 0.37g/l Co 1.37g/l Fe

117 ml/min 0.1M H2SO4

36 ml/min 1M H2SO4

Figure 5-4: Mixer-settler configuration when stripping an organic phase with a lower metal loading (process conditions and results are shown in Table A-25 in Appendix A)

Ni and Co stripping (1st stage) Fe stripping (2nd stage) A/O = 2.3 A/O = 1 Residence time in mixers = 3.8min Residence time in mixers = 6.4min Residence time in settlers = 16.2min Residence time in settlers = 27min

Settler Settler Settler Settler 1 2 3 4 Mixer Mixer Mixer Mixer 50 ml/min 1 2 3 4 3.07g/l Ni 0.96g/l Co 2.5g/l Fe

117 ml/min 0.1M H2SO4

50 ml/min 1M H2SO4

Figure 5-5: Mixer-settler configuration when stripping an organic phase with a higher metal loading (process conditions and results are shown in Table A-26 in Appendix A)

In all tests a 20V% Cyanex 272 solution in Shellsol D70 was used while the temperature in each stage was controlled between 40°C and 50°C by adjusting the heating bath temperature.

Figures 5-2 and 5-3 reveal the configuration that were used for the extraction tests. The pump capacities did not allow for the high O/A ratio of 25 in the pre-loading section (maximum is (200 ml/min)/(20 ml/min) = 10). Consequently only the designed process conditions for the extraction section were considered. Lower nickel aqueous concentrations and O/A ratios were used for the pre-loading section to generate the required nickel loading for the extraction section. For the first extraction test in Figure 5-2 the O/A ratio of 1, 40 g/l NaOH flow rate of 15 ml/min and nickel aqueous feed concentration of

Stellenbosch University http://scholar.sun.ac.za Continuous tests

10.74 g/l in the pre-loading stage were chosen based on the previous batch tests to produce a nickel loading ranging between 3-3.5 g/l. For the extraction section a higher iron concentration (3.6 instead of 3 g/l) was used in the pregnant solution to observe if the generated loading was still adequate at the selected O/A ratio of 1.5 to produce a pure nickel product solution as well as to see if the model could predict the variation regarding this metal inlet concentration. In the second extraction test (Figure 5-3) a lower O/A ratio of 0.67, a higher 40 g/l NaOH flow rate of 18 ml/min and a higher nickel aqueous feed concentration of 15 g/l were chosen to see if generating a higher nickel loading at the same O/A ratio of 1.5 in the extraction section would result in a purer nickel product solution. In the second extraction test the iron concentration was lowered to 3.07 g/l while the cobalt and nickel concentrations remained between 1.02-1.05 g/l and 80-82 g/l, respectively.

Referring to Figures 5-4 and 5-5, higher and lower organic loadings relative to those that were selected in the simulation sheet (2.89 g/l Ni, 0.63 g/l Co and 2 g/l Fe) were used. This was to see if the selected

A/O ratio and H2SO4 concentration used in the simulation was still sufficient to selectively strip cobalt and nickel as well as to see whether the model allows for variations in metal loading. For the stripping test in Figure 5-4 organic loadings of 0.94 g/l Ni, 0.37 g/l Co and 1.37 g/l Fe were stripped with a H2SO4 concentration of 0.1 M and an A/O ratio of 2.3 at the first stage and a H2SO4 concentration of 1 M and an A/O ratio of 0.72 at the second stage. These A/O ratios that were measured compared well with those that were selected in the simulation sheet of 2.25 and 0.75, respectively. The high A/O ratio of 2.3 at the first stage demanded a high acid flow rate resulting in a residence time of 1.9 minutes which is much less than the 10 minutes allowed for the batch tests to reach equilibrium. In order to increase this to a total mixing time of 3.8 minutes, the organic and aqueous product from the first settler were connected to the second mixer as indicated in Figure 5-4. An extra mixer was not added to the second stripping stage since the residence time was already 3.7 minutes. For the second stripping test in Figure 5-5 the organic loading was increased to 3 g/l Ni, 0.96 g/l Co and 2.5 g/l Fe. A second mixer was also added to the second stage to increase the residence time at this stage as shown in Figure 5-5. A higher A/O ratio of 1 was also used to accommodate for the fact that equilibrium might not be reached and try to strip all the metal still left in the organic phase.

5.4 EXPERIMENTAL PROCEDURE

All stock solutions were prepared in the same manner as those in the batch tests. Metal aqueous concentrations were determined via atomic absorption spectrophotometry and the metal loadings were accordingly calculated via a mass balance. The following general start-up, operating and shut-down procedure was used for both the extraction and stripping tests in this specific order.

5.4.1 START UP PROCEDURE

 The heating baths were turned on by adjusting the knob of the heaters to the desired temperature.  The system was then configured as per the test design.  All the fittings were firmly fastened (finger tight).  The position of the impellers was adjusted to the centre of the mixers and close to the bottom without touching the bottom or the sides.

Stellenbosch University http://scholar.sun.ac.za Continuous tests

 All the chucks of the impellers were securely tightened.  A container filled with demineralised water was placed in the heating bath (approximately 5 l). This was used during start-up.  Next the pre-loading, pregnant leach, stripping and NaOH stock solutions were prepared and also placed in the heating baths.  The first stage settler and mixer were then filled half full with the nickel aqueous pre-loading solution and the remaining mixers and settlers was filled half full with the demineralised water (This was only done for the extraction tests, the first stage was also filled with water for the stripping configurations).  10 ml samples of each stock solution were then taken to confirm the metal content in each batch via AAS.  The organic stock solutions were then prepared and all the remaining settlers and mixers were filled to their overflow weirs.  The remaining organic solution was then also placed in the heating bath.  Next the NaOH feed chamber was filled with the NaOH stock solution (only applicable to the extraction configurations).  The temperature and pH sensors were then turned on and the corresponding probes were placed in the mixer and aqueous overflow weirs respectively.  The 1st stage aqueous feed and product tubes were then placed in the nickel pre-loading inventory so that the aqueous product from the 1st stage is recycled back to the pre-loading inventory (only applicable to the extraction configurations).  Next the aqueous feed and product tubes for the rest of the stages were placed in the container filled with demineralised water so that these aqueous product streams are also recycled back to this inventory.  The organic feed tubing for the first stage as well as the organic product tube from the final stage was then placed in the organic stock solution so that the organic phase is also recycled back to the organic inventory. Initially pure and not loaded organic was used for the stripping tests.  The impellers were then turned on and the speed was adjusted to 1000 rpm.  The knobs on the feed pumps were then adjusted to the desired flow rates and they were all activated simultaneously.  10 ml samples after 5 minutes from each mixer were taken to determine the O/A ratio in each stage and the pumping rates were adjusted accordingly to obtain the desired O/A ratio.  A sample from the 1st stage aqueous product was taken to confirm that no loading has occurred yet (only applicable to the extraction configurations).  The liquid interface at each settler was examined and had to remain constant to ensure that the system was at steady state (the aqueous overflow flow weir was shifted up and down to shift the liquid interface up and down in the settler).  All the product tubes were then placed in the product containers and the feed tubing from the water feed as well as the organic feed tube in the loaded organic inventory for the stripping tests were placed in their respective inventory feeds.  The valve for the NaOH feed to the 1st stage was then adjusted to the desired flow rate as shown on the flow metre (only applicable to the extraction configurations).

Stellenbosch University http://scholar.sun.ac.za Continuous tests

5.4.2 OPERATING PROCEDURE

 During operation it was important to ensure that the NaOH inventory remained full by consistently adding NaOH since the container only had a volume of 500 ml (only applicable to the extraction configurations).  A sample was taken every 10 minutes from each aqueous product stream.  The pH, temperature and O/A were also regulated every 10 minutes.

5.4.3 SHUT DOWN PROCEDURE

 During shut down the feed pumps were switched off first.  The impellers were then turned off.  The organic and aqueous phases were then drained from the system.  The mixers, settlers, over-flow weirs, and tubing were rinsed with water.  The mixers and pumps were cleaned with a mild soap or detergent solution).

5.5 RESULTS

5.5.1 EXTRACTION

The nickel loading, sodium concentration in the aqueous product solution and pH for the pre-loading stage for the two configurations in Figures 5-2 and 5-3 are shown in Figures 5-6 (a) and 5-7 (a), respectively. Nickel, cobalt and iron extraction as well as sodium addition to the purified nickel product solution and pH for the extraction section for the two configurations in Figures 5-2 and 5-3 are shown in Figures 5-6 (b) and 5-7 (b), respectively. Zero minutes represents the time after the NaOH valve was opened.

7 100 6 6 90

80 5 5 70 4

4 (g/l)

60 aq (g/l) and pH and (g/l) 3 50 3

aq 2 40 2 1 30 20 0 10 1 -1 and Na pH 0 0

loading, Na loading, -2 -10 -

Ni, Co and Fe extraction Fe (%)Coextraction and Ni, -20 -1

Ni -3 0 20 40 60 80 100 0 20 40 60 80 100 Time (min) Time (min) (a) (b)

10090 10 607080 20304050 0 a 100 -2010 -10 … … Ni Co Fe Na pH q Fe Co Ni, Ni, pH pH e Na (g and and 0 20 40 60 80 100 and Time (min) …

Figure 5-6: Results for pre-loading (1st stage (a)) and extraction (2 nd stage (b)) when generating a lower nickel loading on the organic phase (process conditions and results are shown in Table A-23 in Appendix A)

Stellenbosch University http://scholar.sun.ac.za Continuous tests

7 100 6 6 90

5 5 80 4 70 4 (g/l) 3 60 aq (g/l) and pH and (g/l) 2 50 3 aq 1 40 2 0 30 -1 20 1

-2 10 and Na pH -3 0

loading, Na loading, 0 - -4 Ni, Co and Fe extraction (%) extraction Fe Co and Ni, -10 -1 Ni -5 0 20 40 60 80 100 0 20 40 60 80 100 Time (min) Time (min) (a) (b)

10090 10 607080 20304050 0 a 100 -2010 -10 … … Ni Co Fe Na pH q Fe Co Ni, Ni, pH pH e Na (g and and 0 20 40 60 80 100 and Time (min) … Figure 5-7: Results for pre-loading (1st stage (a)) and extraction (2nd stage (b)) when generating a higher nickel loading on the organic phase (process conditions and results are shown in Table A-24 in Appendix A)

During all the extraction tests no phase separation problems were encountered and a third phase barely even formed in the settlers. Referring to Figure 5-6 (a) there are peaks at 15 minutes in the nickel loading and sodium concentration in the aqueous product solution of 5.8 g/l and 4.48 g/l, respectively. This was caused by a momentarily increase in the NaOH flow rate between 0 and 10 minutes. For the time period between 15 and 25 minutes (Figure 5-6 (a, b)) the pH probes were removed from the aqueous product solutions so that this sudden increase in pH did not interfere with the probes calibration (this is why there are no pH data for this time period).

The negative nickel loading between 0-5 minutes in Figures 5-6 (a) and 5-7 (a) can be explained by referring to equation 35 that was used to calculate nickel loading.

[35]

where: Niorg = Nickel concentration in the organic phase at equilibrium/nickel loading (g/l)

Nioaq = Nickel concentration in the aqueous feed (g/l)

Niaq = Nickel concentration in the aqueous product (g/l)

Faq = Aqueous flow rate nickel pre-loading solution (ml/min)

FNaOH = NaOH flow rate (ml/min)

During the start-up procedure nickel is initially circulated with the organic phase in the first stage without opening the NaOH valve. Consequently five minutes after the NaOH has been opened the effect of the extra volume added by the NaOH stream to the aqueous product stream has not yet been detected in the sample and the loading should in fact be zero by setting FNaOH equal to zero in equation 35. The nickel loading at steady state is however of primary interest therefore equation 35 was still used in this regard irrespective of the time period after the NaOH valve has been opened.

Stellenbosch University http://scholar.sun.ac.za Continuous tests

The initial high extraction for nickel, cobalt and iron in Figures 5-6 (b) and 5-7 (b) is due to the fact that the aqueous product from this stage is still largely water that was used during the start-up procedure before the NaOH valve was opened. Hence it may appear as if nickel, cobalt and iron have been extracted, but this is only because they have not yet made their way through the mixer and settler. Nickel extraction can be used as indication to see when all the demineralised water has been removed from the system (this will typically be when it approaches zero).

From Figure 5-6, steady state was reached approximately 85 minutes after the NaOH valve was opened since al the curves are reaching or are already at a consistent value. The steady state nickel loading ranges between 3.26 and 3.36 g/l. This is close to the loading of 3.2 g/l selected as the optimum loading at an O/A ratio of 1.5 for the extraction section. The pH and cobalt and iron extraction at steady state, range between 4.97-5.05, 94-95% and 97-98% respectively. At nickel loadings ranging between 3.26- 3.36 g/l and an O/A ratio of 1.5, the simulation sheet returns a pH and cobalt and iron extraction range of 4.62-4.67, 95-96% and 99-99%, respectively, which compares well with the experimental results. The nickel loading and O/A ratio were adequate to remove the extra iron added to the pregnant nickel solution (3.6 g/l instead of 3 g/l) as predicted and proven by the simulation sheet and experimental results, respectively.

As mentioned previously the second extraction test was conducted to see if generating a higher nickel loading results in a purer nickel product (in other words if higher cobalt and iron extraction is achieved). Referring to Figure 5-7, steady state was reached after approximately 55 minutes. At steady state the nickel loading ranges between 4.98 and 5.16 g/l and the pH and cobalt and iron extraction between 4.89-4.91, 95-96% and 98-99%, respectively. At the same nickel loading and O/A ratio the simulation sheet returns a pH and cobalt and iron extraction range of 5.21-5.24, 97-98% and 99-100% respectively. Even though there is a slight increase in cobalt and iron extraction as predicted and proven by the second test and simulation, this effect is not severe enough to increase the nickel loading for a purer nickel product. The model under predicts the extraction pH for the lower nickel loading and over predicts the extraction pH at the higher nickel loading whereas it slightly over predicts extraction for both the low and high loading.

The sodium addition to the purified nickel product solution ranges between -0.01-0 g/l and 0.1-0.4 for the low and high nickel loading tests, respectively. The negative addition for the low nickel loading case is due to impure sodium that was in the stock solutions and was subtracted from the concentration given by the AAS (in other words some of the sodium was also extracted). It is also observed that the sodium addition increases as the nickel loading increases. This also justifies why it would be better to rather use the lower loading since the negligible extra extraction obtained from the higher nickel loading comes at the cost of having more sodium in the purified nickel product stream.

5.5.2 STRIPPING

The percentage nickel, cobalt and iron stripped in both stages for the two configurations in Figures 5-4 and 5-5 are shown in Figures 5-8 and 5-9, respectively. Zero minutes represents the time after each feed

Stellenbosch University http://scholar.sun.ac.za Continuous tests

tube is switched from the water and pure organic inventory to the stripping and loaded organic inventory solutions.

100 120 110 90 100 80

90 70 (%) (%) 80 70 60 60 50 50 40 40 tripped S Stripped Stripped 30 30 20 20 10 10 0 0 0 20 40 60 80 100 0 20 40 60 80 100 Time (min) Time (min) (a) (b)

1001101208090 102030405060700 … ri Time (min) … St Ni Co Fe p

100 100 90 90 80 80

70 70 (%) 60 (%) 60 50 50 40 40 tripped tripped tripped tripped S 30 S 30 20 20 10 10 0 0 0 20 40 60 80 100 0 20 40 60 80 100

Time (min) Time (min) (a) (b)

1001101201020304050607080900 … ri

St Ni Co Fe p 0 20 40 60 80 100 Time (min) …

The percentage stripped in both stages was calculated with respect to the metal loading being fed to the first stage. This reveals what percentage of the metal fed to the stripping section ends up in which product stream.

For the stripping tests there were also no phase separation problems with the two phases almost completely disengaging as they flow from the mixer over flow weir to the settler where a third phase was hardly observed. The percentage nickel stripped at 5 minutes after start-up was calculated to be 366.41%. This is not indicated in Figure 5-8 (a) as the y-axis has been reduced to 120% to give a better visual indication of the rest of the data. A possible cause for this unrealistic value is the fact that some of the aqueous phase was still suspended in the loaded organic phase after the two phases were separated

Stellenbosch University http://scholar.sun.ac.za Continuous tests

and could not visually be observed. This then descended to the bottom of the container after a certain time period and was pumped by the organic pump to the 1st stage mixer. Most of the cobalt and iron were extracted while preparing this batch but not all of the nickel; hence this only affected the percentage nickel stripped in Figure 5-8 (a). This problem did not occur for the second stripping test as indicated in Figure 5-9 (a).

Figure 5-8 shows the results for the scenario where the organic phase with the lower metal loading of 0.94 g/l Ni, 0.37 g/l Co and 1.37 g/l Fe was stripped. In the first stage steady state with respect to nickel and cobalt stripping was reached approximately after 45 minutes. Steady state is however not reached with respect to iron stripping. This is also the case in the second stage where iron stripping increases and does not become consistent like nickel and cobalt. From 65 to 85 minutes the percentage of nickel, cobalt and iron stripped in the 1st stage ranges between 97-101%, 74-78% and 11-5% and in the 2nd stage between 4-5%, 7-9% and 27-34%, respectively. The simulated percentage nickel, cobalt and iron stripped for this metal loading and process conditions as shown in Figure 5-4 for the first stage are 99.98%, 99.95% and 0.28% and for the second stage are 0.02%, 0.05% and 99.45%, respectively. The almost ideal separation of nickel and cobalt from iron predicted by the models is not reflected in the experimental results. Referring to Figure 3-18 a mixing time of approximately 410 seconds or 6.83 minutes is required for equilibrium. This is almost double the residence time of 3.8 and 3.7 minutes used for the 1st and 2nd stage mixers, respectively. The simulated and experimental results with respect to nickel stripping corresponds the best. A likely explanation, when considering the stripping kinetics in figure 3-18, is the fact that nickel is stripped faster than either cobalt or iron with more than half the nickel being stripped after 60 seconds. The opposite can be applied with respect to iron stripping, since the stripping kinetics in this regard are slower than nickel or cobalt and therefore the residence time of 3.7 minutes is not enough to reach equilibrium in the second stage. This is reflected in the low percentage of iron stripped as opposed to the high stripping predicted by the model.

Next the stripping results of the higher metal loading configuration in Figure 5.5 (3.07 g/l Ni, 0.95 g/l Co and 0.95 g/l Fe) will be considered. Due to the poor iron stripping obtained (maximum of 33.72%) in the 2nd stage in Figure 5.8 an extra mixer and settler were added to the 2nd stage as discussed previously to increase the residence time. This adjustment increased the 2nd stage mixer residence time to 6.4 minutes while the 1st stage mixer residence time remained at 3.8 minutes. A higher A/O ratio of 1 was also used with the aim of increasing iron extraction at this stage. Referring to Figure 5.9, steady state was reached approximately after 85 minutes which was the duration for the previous stripping test. Consequently larger inventories were prepared for the second stripping test which successfully yielded steady state. At steady state the percentage nickel, cobalt and iron stripped in the 1st stage ranges between 88-90%, 57-60% and 20-23% and in the 2nd stage between 16-19%, 31-32% and 69-70%, respectively. The simulated percentage nickel, cobalt and iron stripped for the first stage is 99.98%, 99.95% and 0.28% and for the second stage is 0.02%, 0.05% and 99.48%, respectively. The simulated results once again predict excellent separation of nickel and cobalt from iron which is not the case for the experimental results. The additional nickel and cobalt loaded onto the organic phase decreased the nickel and cobalt stripping by approximately 10% in the 1st stage while the remainder ended up in the second stage aqueous product stream. As expected, the increased residence time and A/O increased iron stripping in the second stage even though a higher loading was used. Complete stripping of the organic phase was however again not attained.

Stellenbosch University http://scholar.sun.ac.za Continuous tests

It is clear that the simulation sheet made more accurate predictions for the extraction tests than for the stripping tests. This is probably due to the fact that the extraction tests came much closer to equilibrium than the stripping tests did. Even though the stripping tests did not generate the favourable results the extraction tests did, it is still speculated that selective stripping of cobalt and nickel from iron can be achieved by doing some fine tuning with respect to the A/O ratio and H2SO4 concentration at each stripping stage as well as increasing the residence time for iron stripping.

Stellenbosch University http://scholar.sun.ac.za Conclusions

CHAPTER 6

6 CONCLUSIONS

The following conclusions are made from the literature review, batch tests, modelling and simulations and continuous tests. Recommendations for future work are also given.

6.1 LITERATURE REVIEW

Solvent extraction with Cyanex 272 as extracting reagent is the most popular separation process to remove cobalt and iron from pregnant nickel leach solutions. This is based on the extensive amount of research accompanied with favourable results that are available as well as for the fact that it is currently being applied in industry specifically for selective extraction of cobalt from nickel (Flett, 2005).

Most research focuses on either using saponified Cyanex 272 or pure Cyanex 272 during the extraction process. This however contaminates the aqueous product solution with sodium. A limited amount of research is available regarding the use of nickel salts of Cyanex 272 to extract cobalt and iron from nickel sulfate solutions which eliminates any sodium additions to the purified nickel aqueous product solution.

Equilibrium pH/NaOH addition, A/O feed ratio, extractant concentration, temperature, agitation time, nickel loading onto the organic and acid concentration are the most important operating variables in solvent extraction for this system.

6.2 BATCH TESTS

The pre-loading, extraction and stripping experimental setup are considered as being highly repeatable with a minimum r-value of 0.96 for the three replicates for all the factorial designs. The only exception was with nickel extraction during the extraction factorial design which yielded a r-value of 0.371. This was due to the fact that nickel extraction remained between 0-10% and is consequently hard to reproduce like cobalt and iron extraction since the mean square within groups do not differ significantly from the mean square between groups.

The pre-loading kinetics are the fastest followed by extraction with pure Cyanex 272, then extraction with the nickel salts of Cyanex 272 and then stripping, with equilibrium roughly being reached within 20, 100, 150 and 410 seconds, respectively.

Provided there is enough nickel available in the aqueous phase, nickel loading of Cyanex 272 is primarily a function of the NaOH addition and not the initial nickel aqueous concentration. In other words nickel

Stellenbosch University http://scholar.sun.ac.za Conclusions

loading of Cyanex 272 is a function of the quantity of OH- ions that can be used to neutralise the H+ ions being exchanged for nickel in the aqueous phase; hence using higher initial nickel aqueous concentrations will not generate higher loadings.

When comparing pure Cyanex 272 and the nickel salt of Cyanex 272 to remove 1 g/l Co and 3 g/l Fe from an 80 g/l Ni solution, 4 g/l and 0.05 g/l of sodium ends up in the purified nickel product streams, respectively. Hence using the nickel salt of Cyanex 272 is the more attractive approach if minimization of sodium contamination in the product solution is of concern.

Increasing the extractant concentration decreases the effect the diluents have on the extracting power of Cyanex 272. These effects were however not large enough, even at the low concentrations, and Shellsol D70 was chosen as diluent for subsequent experimental work since it is one of diluents currently being manufactured for the extraction of cobalt and nickel with Cyanex 272 by Shell Chemicals in South Africa.

From the pre-loading factorial design, O/A ratio and NaOH addition are the two dominating factors with respect to nickel pre-loading. Temperature and extractant concentration on the other hand do not influence nickel loading to a significant extent. The equilibrium pH at which loading occurs decreases as the extractant concentration increases caused by the excess H+ being added to the dispersion.

According to the factorial design for the extraction section, O/A ratio and nickel loading are the most influential factors with respect to cobalt and iron extraction. Preliminary experiments revealed that increasing the extractant concentration from 10 V% to 20 V% increases cobalt extraction but has no significant effect on iron extraction. According to the factorial design extractant concentration does not influence either cobalt or iron extraction. Consequently a 20V% Cyanex 272 solution was chosen for subsequent tests since it increases cobalt extraction and decreases possible ligand overloading leading to phase separation problems. Temperature influences cobalt and iron extraction at nickel loadings lower than approximately 4 g/l and an O/A ratio of 1 with cobalt and iron extraction decreasing and increasing, respectively, as temperature increases. At nickel loadings higher than 4 g/l for an O/A ratio of 1 temperature does not have a significant influence on cobalt and iron extraction. A temperature of 50°C was chosen for the pre-loading and extraction sections since it minimizes phase separation problems and nickel loadings and O/A ratios close to 4 g/l and 1, respectively, are recommended.

The nickel extraction data with Ni-Cyanex 272 was scattered as a function of nickel loading and O/A ratio but consistently remained within -6% and 6% with a single exception of 8.77% at a nickel loading of 0.56 g/l and an O/A ratio of 2. The sodium addition to the purified nickel product solution increases as the O/A ratio and nickel loading increases. This is caused by the fact that higher nickel loadings demand higher sodium additions in the pre-loading section which increases the probability of sodium uptake by the organic phase and which consequently ends up in the purified nickel stream.

In essence all of the (99-100%) iron is removed from the pregnant solution for the selected nickel loading and O/A ratio (1.5 and 3.2 g/l respectively). A small quantity of cobalt however still remains after extraction (95-96%) even at the maximum nickel loadings and O/A ratios tested.

Stellenbosch University http://scholar.sun.ac.za Conclusions

From the nickel loading extraction isotherms, one theoretical stage is calculated to generate the required loading (3.2 g/l) at a pH of 7 and an O/A ratio of 25 from the 80 g/l purified nickel solution.

A/O ratio and acid concentration are the most influential factors with respect to stripping. The metal loading on the organic phase also influences stripping and as a result different metal loadings will require different A/O ratios and stripping agent concentrations. Temperature only influences nickel stripping. This effect is however not large and a temperature of 50°C was used for the subsequent stripping tests as well.

Stripping occurs at a lower pH than extraction; hence the metal complexes are more stable in the organic phase than in the aqueous phase.

Selective stripping of nickel and cobalt are possible to a certain extent but are limited by phase separation problems and hence not practical. Selective stripping of nickel and cobalt from iron is however possible and do not present any phase separation problems.

A 0.1 M H2SO4 solution at an A/O ratio of 2.25 is recommended to strip cobalt and nickel and a 1 M

H2SO4 solution at an A/O ratio of 0.75 to strip the remaining iron from the organic phase.

6.3 MODELLING AND SIMULATION

The model derived from the equilibrium constant for nickel loading as a function of NaOH/Nio and O/A ratio gave a good fit with a R2 value of 0.97.

The sigmoid function was also successfully manipulated to model the loading pH as a function of NaOH/Nio and O/A ratio, the concentration of cobalt and iron in the extraction aqueous product solution and extraction pH as a function of nickel loading and O/A ratio and nickel, cobalt and iron concentration in the aqueous product strip solution as a function of H2SO4 concentration and A/O ratio. The lowest R2 value out of all these fits with the modified sigmoid function was 0.94 for iron concentration in the stripped product solution and the extraction pH which was still regarded as a good model fit.

Using these models and mass-balance calculations, a simulation sheet was successfully developed to predict all the aqueous and organic product concentrations (except for nickel in the extraction product solutions which can be selected arbitrarily) as well as the extraction or stripping at each stage and pH for the pre-loading and extraction stages. The input values that can be varied include the metal concentrations and flow rate of the pregnant solution, the O/A ratio for the pre-loading and extraction sections, A/O ratio for the stripping sections, the relative moles of NaOH being added with respect to nickel in the aqueous feed (NaOH/Nio), the NaOH concentration used for the pre-loading section and the H2SO4 concentration of the two streams being fed to the stripping sections

Stellenbosch University http://scholar.sun.ac.za Conclusions

After conducting a sensitivity analysis with the simulation sheet, it was concluded that the selected operating variables for the extraction circuit are robust towards variations in the feed concentration for the extraction and stripping sections.

6.4 CONTINUOUS TESTS

The mixer-settler setup was successfully used to extract cobalt and iron from the pregnant nickel solution by using the pre-loading procedure.

The predictions made regarding cobalt extraction, iron extraction and extraction pH by the simulation sheet were accurately re-produced during the two extraction tests.

The O/A ratio of 1.5 and nickel loading ranging between 3.26 g/l and 3.36 g/l at steady state, which is close to the 3.2 g/l selected as the optimum during the batch tests, were adequate to remove the additional iron (3.6 g/l instead of 3 g/l) and cobalt (1 g/l) from the concentrated nickel solution.

Increasing the nickel loading so that it ranges between 4.98 g/l and 5.16 g/l at steady state did not have a significant influence on cobalt and iron extraction (even though the iron concentration was decreased to 3 g/l in the pregnant solution). This increase in nickel loading also increased the sodium addition to the purified nickel solution from -0.01-0 g/l to 0.1-0.4 g/l and it was concluded that the lower loading is still the better option.

The stripping results did not accurately reflect the predictions made by the simulations. A possible cause is that equilibrium was not reached in each stage as the pump capacities did not allow for low enough flow rates. This statement was justified during the second stripping test where iron extraction increased as the residence time increased by adding an additional mixer and settler to the second stripping stage.

Even though the selected H2SO4 concentration and A/O ratio of 0.1 M and 2.25, respectively, to selectively strip nickel and cobalt from iron did not completely achieve this goal, it is believed that this is still possible through some fine tuning of these two variables. The same goes for the failed attempt to regenerate the organic phase and completely strip iron with 1 M H2SO4 at an A/O ratio of 0.75 and 1 (increasing the residence time will also increase iron extraction).

6.5 FUTURE WORK

The following comments are made regarding future work:

 Modify the mixer-settler setup to test the simulation sheet with respect to pre-loading the organic with 80 g/l Ni.

Stellenbosch University http://scholar.sun.ac.za Conclusions

 Use the mixer-settler setup to find the optimum conditions (H2SO4 concentration and A/O ratio) to selectively strip nickel and cobalt from iron as well as to completely remove iron after nickel and cobalt have been stripped.  Conduct a fundamental study regarding the phase separation problems (single emulsion that formed) that were encountered during the stripping tests by considering factors like viscosity and liquid interface tension.  Conduct a kinetic study regarding the reaction mechanism by which Ni-Cyanex 272 removes cobalt and iron from the concentrated nickel sulfate solution. This will provide a more fundamental understanding of the behaviour of nickel, cobalt and iron in the pregnant solution during extraction.

Stellenbosch University http://scholar.sun.ac.za References

REFERENCES

Agatzini-Leonardou, S, Tsakiridis, P, Oustadakis, P, Karidakis, Katsiapi, A (2009), "Hydrometallurgical process for the separation and recovery of nickel from sulfate heap leach liquor of nickeliferrous laterite ores", Minerals Engineering, vol. 22, no. 14, pp. 1181-1192.

Ajgaonkar, H and Dhadke, P (1997), "Solvent extraction separation of iron (III) and aluminium (III) from other elements with Cyanex 302", Talanta, vol. 44, no. 4, pp. 563-570.

Bhaskara Sarma, P and Reddy, B (2002), "Liquid-liquid extraction of nickel at macro-level concentration from sulfate/chloride solutions using phosphoric acid based extractants", Minerals Engineering, vol. 15, no. 6, pp. 461-464.

Biswas, R, Habib, M and Karmakar, A (2007), "Kinetics of Solvent Extraction of Iron (III) from Sulfate Medium by Purified Cyanex 272 using a Lewis Cell", Solvent Extraction and Ion Exchange, vol. 25, no. 1, pp. 79-98.

Biswas, R and Singha, H (2006), "Purified Cyanex 272: Its interfacial adsorption and extraction characteristics towards iron (III)", Hydrometallurgy, vol. 82, no. 1-2, pp. 63-74.

Clark, PDA, Cole, PM and Fox, MH (1993), "The purification of nickel sulphate using di-(2-ethylhexyl) phosphoric acid", Proceedings of the International Solvent Extraction conference, vol. 1, pp. 175- 182.

Darvishi, D, Haghshenas, D, Alamdari, EK, Sadrnezhaad, S and Halali, M (2005), "Synergistic effect of Cyanex 272 and Cyanex 302 on separation of cobalt and nickel by D2EHPA", Hydrometallurgy, vol. 77, no. 3-4, pp. 227-238.

Devi, N, Nathsarma, K and Chakravortty, V (1998), "Separation and recovery of cobalt (II) and nickel (II) from sulfate solutions using sodium salts of D2EHPA, PC 88A and Cyanex 272", Hydrometallurgy, vol. 49, no. 1-2, pp. 47-61.

Devi, N, Nathsarma, K and Chakravortty, V (1994), "Sodium salts of D2EHPA, PC-88A and Cyanex-272 and their mixtures as extractants for cobalt (II)", Hydrometallurgy, vol. 34, no. 3, pp. 331-342.

Flett, DS (2005), "Solvent extraction in hydrometallurgy: the role of organophosphorus extractants", Journal of Organometallic Chemistry, vol. 690, no. 10, pp. 2426-2438.

Flett, DS (1972), "A survey of reagents used for solvent extraction in extractive metallurgy ", Proceedings of the International Symposium Solvent Extraction in Metallurgical Processes, vol. 1, pp. 22-30.

Foust, AS, Wenzel, LA, Maus, L, Clump, CW and Andersen, LB (2008), "Solvent extraction", Wiley, India.

Gandhi, M, Deorkar, N and Khopkar, S (1993), "Solvent extraction separation of cobalt (II) from nickel and other metals with Cyanex 272", Talanta(Oxford), vol. 40, no. 10, pp. 1535-1539.

Habashi, F (1999), "A textbook of hydrometallurgy, 2nd eddition", Quabec City, Canada.

Stellenbosch University http://scholar.sun.ac.za References

Hubicki, Z and Hubicka, H (1996), "Studies on the extraction process of nickel (II) sulfate purification using Cyanex 272", Hydrometallurgy, vol. 40, no. 1-2, pp. 65-76.

Kang, J, Senanayake, G, Sohn, J and Shin, S.M (2009), "Recovery of cobalt sulfate from spent lithium ion batteries by reductive leaching and solvent extraction with Cyanex 272", Hydrometallurgy, vol. 40, no. 1-2, pp. 168-171.

Konishi, Y and Takano, K (2002), "Precipitation of nickel hydroxide by hydrolytic stripping of nickel phosphinate dissolved in organic solvent", Ind.Eng.Chem.Res, vol. 41, no. 16, pp. 3999-4003.

Li, L, Xu, S, Ju, Z and Wu, F (2009), "Recovery of Ni, Co and rare earths from spent Ni-metal hydride batteries and preparation of spherical Ni (OH) 2", Hydrometallurgy, vol. 100, no. 1-2, pp. 41-46.

Lindell, E, Jääskeläinen, E, Paatero, E and Nyman, B (2000), "Effect of reversed micelles in Co/Ni separation by Cyanex 272", Hydrometallurgy, vol. 56, no. 3, pp. 337-357.

Lothongkum, A.W, Khemglad, Y, Usomboon, N and Pancharoen, U (2009), "Selective recovery of nickel ions from wastewater of stainless steel industry via HFSLM", Journal of Alloys and Compounds, vol. 476, no. 1-2, pp. 940-949.

Measey, GJ, Silva, JB and Di-Bernardo, M (2003), "Testing for repeatability in measurements of length and mass in Chthonerpeton indistinctum (Amphibia: Gymnophiona), including a novel method of calculating total length of live caecilians", Herpetological Review, vol. 34, no. 1, pp. 35-38.

Mihaylov, I (2003), "Solvent extractants for nickel and cobalt: New opportunities in aqueous processing", JOM Journal of the Minerals, Metals and Materials Society, vol. 55, no. 7, pp. 38-42.

Miralles, N, Sastre, A, Figuerola, E and Martinez, M (1992), "Solvent extraction of iron (III) by bis (2, 4, 4- trimethylpentyl) phosphinic acid: experimental equilibrium study", Hydrometallurgy(Amsterdam), vol. 31, no. 1-2, pp. 1-12.

Mohapatra, D, Hong-In, K, Nam, CW and Park, KH (2007), "Liquid-liquid extraction of aluminium (III) from mixed sulfate solutions using sodium salts of Cyanex 272 and D2EHPA", Separation and Purification Technology, vol. 56, no. 3, pp. 311-318.

Nogueira, CA, Oliveira, PC and Pedrosa, FM (2009), "Separation of cadmium, cobalt, and nickel by solvent extraction using the nickel salts of the extractants", Solvent Extraction and Ion Exchange, vol. 27, no. 2, pp. 295-311.

Nogueira, C, Oliveira, P and Pedrosa, F (2003), "Scrubbing of cadmium and nickel from Cyanex 272 loaded with cobalt", Solvent Extraction and Ion Exchange, vol. 21, no. 5, pp. 717-734.

O'Callaghan, J and Chamberlain, T (2002), "Solvent extraction of impurity metals from a valuable metal sulfate solution", European Patent EP1252345, assigned to WMC Resources Ltd, Perth.

Parhi, P, Panigrahi, S, Sarangi, K and Nathsarma, K (2008), "Separation of cobalt and nickel from ammoniacal sulfate solution using Cyanex 272", Separation and Purification Technology, vol. 59, no. 3, pp. 310-317.

Park, BRRKH (2007), "Process for the Recovery of Cobalt and Nickel from Sulfate Leach Liquors with Saponified Cyanex 272 and D2EHPA", Separation Science and Technology, vol. 42, no. 9.

Stellenbosch University http://scholar.sun.ac.za References

Park, KH and Mohapatra, D (2006), "Process for cobalt separation and recovery in the presence of nickel from sulfate solutions by Cyanex 272", Metals and Materials International, vol. 12, no. 5, pp. 441- 446.

Perry, RH and Green, DW (2007), Perry's chemical engineers' handbook, McGraw-Hill Professional.

Reddy, BR, Priya, DN and Park, KH (2006), "Separation and recovery of cadmium (II), cobalt (II) and nickel (II) from sulfate leach liquors of spent Ni-Cd batteries using phosphorus based extractants", Separation and Purification Technology, vol. 50, no. 2, pp. 161-166.

Richardson, JF, Coulson, JM, Harker, JH and Backhurst, J (2002), Coulson and Richardson's Chemical Engineering: Particle technology and separation processes, Butterworth-Heinemann.

Ritcey, GM, Ashbrook, AW and Lucas, BH (1975),"Development of a solvent extraction process for the separation of cobalt from nickel", CIM Bull, vol 68, no 1, pp 111-123.

Rodrigues, LEOC and Mansur, MB (2009), "Hydrometallurgical separation of rare earth elements, cobalt and nickel from spent nickel-metal-hydride batteries", Journal of Power Sources, vol. 195, no. 11, pp. 3735-3741.

Sahu, S, Agrawal, A, Pandey, B and Kumar, V (2004), "Recovery of copper, nickel and cobalt from the leach liquor of a sulphide concentrate by solvent extraction", Minerals Engineering, vol. 17, no. 7- 8, pp. 949-951.

Sarangi, K, Reddy, B and Das, R (1999a), "Extraction studies of cobalt (II) and nickel (II) from chloride solutions using Na-Cyanex 272. Separation of Co (II)/Ni (II) by the sodium salts of D2EHPA, PC88A and Cyanex 272 and their mixtures", Hydrometallurgy, vol. 52, no. 3, pp. 253-265.

Sarangi, K, Reddy, B and Das, R (1999b), "Extraction studies of cobalt (II) and nickel (II) from chloride solutions using Na-Cyanex 272. Separation of Co (II)/Ni (II) by the sodium salts of D2EHPA, PC88A and Cyanex 272 and their mixtures", Hydrometallurgy, vol. 52, no. 3, pp. 253-265.

Sole, KS, Feather, AM and Cole , PM (2005), "Solvent extraction in southern Africa: An update of some recent hydrometallurgical developments", Hydrometallurgy, vol. 78, no. 1, pp. 52-78.

Swain, B, Jeong, J, Lee, J and Lee, GH (2008), "Development of process flow sheet for recovery of high pure cobalt from sulfate leach liquor of LIB industry waste: A mathematical model correlation to predict optimum operational conditions", Separation and Purification Technology, vol. 63, no. 2, pp. 360-369.

Swain, B, Jeong, J, Lee, J and Lee, GH (2006), "Separation of cobalt and lithium from mixed sulfate solution using Na-Cyanex 272", Hydrometallurgy, vol. 84, no. 3-4, pp. 130-138.

Tait, B (1993), "Cobalt-nickel separation: the extraction of cobalt (II) and nickel (II) by Cyanex 301, Cyanex 302 and Cyanex 272", Hydrometallurgy(Amsterdam), vol. 32, no. 3, pp. 365-372.

Tsakiridis, P and Agatzini-Leonardou, S (2004), "Process for the recovery of cobalt and nickel in the presence of magnesium from sulfate solutions by Cyanex 272 and Cyanex 302", Minerals Engineering, vol. 17, no. 7-8, pp. 913-923.

Stellenbosch University http://scholar.sun.ac.za References

Van de Voorde, I, Latruwe, K, Pinoy, L, Courtijn, E and Verpoort, F (2007), "Complexation Behavior of Iron (III) with Aloxime 800, D 2 EHPA, CYANEX 272, CYANEX 302, and CYANEX 301", Solvent Extraction and Ion Exchange, vol. 25, no. 6, pp. 809-829.

Van de Voorde, I, Pinoy, L, Courtijn, E and Verpoort, F (2006), "Equilibrium Studies of Nickel (II), Copper (II), and Cobalt (II) Extraction with Aloxime 800, D 2 EHPA, and Cyanex Reagents", Solvent Extraction and Ion Exchange, vol. 24, no. 6, pp. 893-914.

Yu, X, Gu, G, Fu, X, Su, B and Zhang, H (2001), "Thermochemistry study on the extraction of Co (II) and Ni (II) with Cyanex 272 in kerosene", Solvent Extraction and Ion Exchange, vol. 19, no. 5, pp. 939-943.

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

APPENDIX A (EXPERIMENTAL AND MODELLING RESULTS)

A1 BATCH TESTS

Table A-1: Extraction kinetics of nickel, cobalt and iron with pure Cyanex 272

Diluent: Kerosene Cyanex 272 (V%): 20 Temperature (°C): 50 3 M NaOH (ml): 1 O/A (ml/ml): 20/20

Nio Coo Feo Ni Co Fe Ni Co Fe Agitation pH aqueous aqueous aqueous aqueous aqueous aqueous extracted extracted extracted period (s) (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (%) (%) (%) 20 6.15 88 1.05 3.3 76 0.611 1.022 9.32 38.90 67.48 40 5.95 88 1.05 3.3 78 0.378 0.468 6.93 62.20 85.11 60 5.5 88 1.05 3.3 80 0.212 0.219 4.55 78.80 93.03 100 4.87 88 1.05 3.3 80 0.143 0.021 4.55 85.70 99.33 150 4.65 88 1.05 3.3 80 0.162 0.014 4.55 83.80 99.55 300 4.68 88 1.05 3.3 82 0.119 0.022 2.16 88.10 99.30

Table A-2: Pre-loading kinetics of Cyanex 272 with nickel

Diluent: Kerosene Cyanex 272 (V%): 20 6.5.1 6.5.2 6.5.3 Temperature (°C): 50 6.5.4 6.5.5 6.5.6 O/A (ml/ml): 20/20 6.5.7

3 M NaOH (ml) Agitation period (s) Nio aqueous (g/l) Ni aqueous (g/l) Ni extracted (%) 1 20 10.75 6.1 40.42 1 40 10.75 6.3 38.47 1 60 10.75 6.2 39.44 1 80 10.75 6.1 40.42 1 100 10.75 6 41.40 1 150 10.75 5.8 43.35 1 200 10.75 6.2 39.44 1 250 10.75 6.3 38.47 1 300 10.75 6.2 39.44 2 20 10.75 2.1 78.51 2 40 10.75 1.9 80.56 2 60 10.75 2 79.53 2 80 10.75 2.1 78.51 2 100 10.75 2 79.53 2 150 10.75 1.8 81.58 2 200 10.75 2 79.53 2 250 10.75 2 79.53 2 300 10.75 1.9 80.56

Stellenbosch University http://scholar.sun.ac.za

Appendix A (Experimental and Modelling results) Appendix A (Experimental and Modelling results) (Experimental and Modeling results)

Table A-3: Extraction kinetics of nickel, cobalt and iron with the nickel salt of Cyanex 272

Cyanex 272 (V%): 20 Temperature (°C): 50 6.5.8 6.5.9 6.5.10 6.5.11 6.5.12 6.5.13 6.5.14 6.5.15 6.5.16 O/A (ml/ml): 20/20

Nickel loading Agitation Nio aqueous Coo aqueous Feo aqueous Ni aqueous Co aqueous Fe aqueous Co extracted Fe extracted Diluent Ni extracted (%) (g/l) period (s) (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (%) (%) 2.47 shellsol D70 20 83.30 1.07 2.88 79.10 0.66 1.41 5.04 38.50 51.22 2.47 shellsol D70 40 83.30 1.07 2.88 85.90 0.65 1.39 -3.12 39.44 51.74 2.47 shellsol D70 60 83.30 1.07 2.88 80.80 0.70 1.29 3.00 34.21 55.21 2.47 shellsol D70 100 83.30 1.07 2.88 80.70 0.88 1.11 3.12 17.66 61.39 2.47 shellsol D70 150 83.30 1.07 2.88 76.80 1.00 0.83 7.80 6.92 71.22 2.47 shellsol D70 300 83.30 1.07 2.88 79.70 1.01 0.68 4.32 6.07 76.53 4.09 kerosene 20 86.33 1.08 3.30 86.00 0.27 0.52 0.38 75.23 84.39 4.09 kerosene 40 86.33 1.08 3.30 87.00 0.18 0.34 -0.78 83.40 89.58 4.09 kerosene 60 86.33 1.08 3.30 87.00 0.17 0.26 -0.78 84.42 92.03 4.09 kerosene 80 86.33 1.08 3.30 87.00 0.17 0.23 -0.78 84.04 93.00 4.09 kerosene 100 86.33 1.08 3.30 86.00 0.20 0.11 0.38 81.08 96.79 4.09 kerosene 150 86.33 1.08 3.30 88.00 0.29 0.01 -1.93 73.10 99.58 4.09 kerosene 250 86.33 1.08 3.30 88.00 0.30 0.00 -1.93 72.17 99.94 4.09 kerosene 300 86.33 1.08 3.30 88.00 0.31 0.02 -1.93 71.06 99.45 4.66 shellsol D70 20 83.30 1.07 2.88 79.80 0.16 0.20 4.20 85.05 93.09 4.66 shellsol D70 40 83.30 1.07 2.88 79.00 0.11 0.07 5.16 89.44 97.43 4.66 shellsol D70 60 83.30 1.07 2.88 77.50 0.12 0.05 6.96 89.16 98.30 4.66 shellsol D70 100 83.30 1.07 2.88 77.70 0.13 0.02 6.72 88.13 99.44 4.66 shellsol D70 150 83.30 1.07 2.88 76.80 0.15 0.00 7.80 86.45 100.00 4.66 shellsol D70 300 83.30 1.07 2.88 77.60 0.15 0.00 6.84 86.17 100.00 8.48 kerosene 20 86.33 1.08 3.30 93.00 0.02 0.01 -7.73 98.33 99.73 8.48 kerosene 40 86.33 1.08 3.30 92.00 0.01 0.01 -6.57 98.70 99.85 8.48 kerosene 60 86.33 1.08 3.30 88.00 0.02 0.01 -1.93 98.61 99.70 8.48 kerosene 80 86.33 1.08 3.30 90.00 0.02 0.01 -4.25 98.61 99.82 8.48 kerosene 100 86.33 1.08 3.30 92.00 0.02 0.01 -6.57 98.33 99.85 8.48 kerosene 150 86.33 1.08 3.30 88.00 0.02 0.00 -1.93 98.14 99.88 8.48 kerosene 250 86.33 1.08 3.30 87.00 0.02 0.00 -0.78 98.14 99.88 8.48 kerosene 300 86.33 1.08 3.30 87.00 0.02 0.00 -0.78 98.24 99.91

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

Table A-4: Pre-loading of Cyanex 272 with different initial nickel aqueous concentrations

Diluent: Kerosene Temperature (°C): 50 6.5.17 6.5.18 6.5.19 Agitation period (min): 5 6.5.20 6.5.21 6.5.22 O/A (ml/ml): 20/20 6.5.23 6.5.24 6.5.25

Cyanex 272 (V%) 3 M NaOH (ml) Nio aqueous (g/l) Ni aqueous (g/l) Ni extracted (%) Ni-loading (g/l) pH 10 0.25 6.15 5.41 10.93 0.67 5.63 10 0.5 6.15 4.3 28.33 1.74 5.9 10 0.75 6.15 3.48 41.29 2.54 6.33 10 1 6.15 2.15 63.29 3.89 6.5 10 1.3 6.15 1.25 78.35 4.82 6.92 10 1.6 6.15 0.02 99.65 6.13 8.08 10 2 6.15 0 100.00 6.15 8.9 10 0.25 10.5 9.84 5.11 0.54 5.68 10 0.5 10.5 8.48 17.22 1.81 5.99 10 0.75 10.5 7.68 24.11 2.53 6.13 10 1 10.5 6 40.00 4.20 6.69 10 1.3 10.5 5.08 48.47 5.09 6.74 10 1.6 10.5 3.36 65.44 6.87 6.91 10 2 10.5 1.74 81.77 8.59 7.07 10 2.25 10.5 0.73 92.27 9.69 7.23 10 0.25 21.2 19.9 4.77 1.01 6 10 0.5 21.2 18.2 12.20 2.59 6.2 10 0.75 21.2 17.6 13.97 2.96 6.35 10 1 21.2 16.2 19.96 4.23 6.53 10 1.3 21.2 14.9 25.15 5.33 6.55 10 1.6 21.2 12.9 34.18 7.25 6.78 10 2 21.2 11.5 40.12 8.51 6.79 10 2.25 21.2 9.98 47.63 10.10 6.81 20 0.25 6.15 4.99 17.85 1.10 5.47 20 0.5 6.15 4.02 33.00 2.03 5.56 20 0.75 6.15 2.93 50.57 3.11 5.82 20 1 6.15 2.09 64.32 3.96 5.91 20 1.3 6.15 0.75 87.01 5.35 6.23 20 1.6 6.15 0.07 98.77 6.07 6.76 20 2 6.15 0 100.00 6.15 7.72 20 2.25 6.15 0 100.00 6.15 8.06 20 0.25 10.5 8.97 13.50 1.42 5.47 20 0.5 10.5 8.19 20.05 2.11 5.64 20 0.75 10.5 7 30.83 3.24 5.72 20 1 10.5 5.83 41.70 4.38 5.83 20 1.3 10.5 4.34 55.98 5.88 5.95 20 1.6 10.5 3.19 67.19 7.05 6.13 20 2 10.5 1.75 81.67 8.58 6.4 20 2.25 10.5 0.78 91.74 9.63 6.5 20 0.25 21.2 19.1 8.87 1.88 5.22 20 0.5 21.2 18.3 11.52 2.44 5.4 20 0.75 21.2 17.1 16.12 3.42 5.55 20 1 21.2 16.2 19.96 4.23 5.76 20 1.3 21.2 14 29.47 6.25 5.83 20 1.6 21.2 13 33.98 7.20 6.3 20 2 21.2 11.1 42.61 9.03 6.35 20 2.25 21.2 9.56 49.83 10.56 6.65

Stellenbosch University http://scholar.sun.ac.za

Appendix A (Experimental and Modelling results) Appendix A (Experimental and Modelling results) (Experimental and Modeling results)

Table A-5: Sodium addition to nickel product solution as a function of cobalt and iron extraction with the nickel loadings of Cyanex 272 as generated in Table A-4

Diluent: Kerosene Temperature (°C): 50 6.5.26 6.5.27 6.5.28 6.5.29 6.5.30 6.5.31 6.5.32 6.5.33 Agitation period (min): 5 6.5.34 6.5.35 6.5.36 6.5.37 6.5.38 6.5.39 6.5.40 6.5.41 O/A (ml/ml): 20/20 Na addition to aqueous Cyanex 272 Ni-loading Ni aqueous Co aqueous Fe aqueous Na aqueous Ni aqueous Co aqueous Fe aqueous Na aqueous Ni extracted Co extracted Fe extracted o o o o phase after pH (V%) (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (%) (%) (%) extraction (g/l) 10 0.67 85 1.06 3.1 0.26 81 1.03 1.8 0.26 4.71 2.83 41.94 0 2.9 10 1.74 85 1.06 3.1 0.26 79.9 1.02 0.85 0.31 6 3.77 72.58 0.05 3.18 10 2.54 85 1.06 3.1 0.26 83.8 1.03 0.04 0.28 1.41 2.83 98.71 0.02 3.65 10 3.89 85 1.06 3.1 0.26 78.4 0.37 0 0.3 7.76 65.09 100 0.04 4.81 10 4.82 85 1.06 3.1 0.26 80.2 0.22 0 0.3 5.65 79.25 100 0.04 5.24 10 6.13 85 1.06 3.1 0.26 77.5 0.21 0 0.57 8.82 80.19 100 0.31 5.86 10 6.15 85 1.06 3.1 0.26 74.5 0.22 0 1.14 12.35 79.25 100 0.88 6.4 10 0.54 85 1.06 3.1 0.26 78.2 1.01 1.81 0.23 8 4.72 41.61 -0.03 3.12 10 1.81 85 1.06 3.1 0.26 81 1.01 0.87 0.26 4.71 4.72 71.94 0 3.16 10 2.53 85 1.06 3.1 0.26 84.4 0.93 0.05 0.28 0.71 12.26 98.39 0.02 3.91 10 4.2 85 1.06 3.1 0.26 87.4 0.36 0 0.25 -2.82 66.04 100 -0.01 4.85 10 5.09 85 1.06 3.1 0.26 85.9 0.22 0 0.33 -1.06 79.25 100 0.07 5.3 10 6.87 85 1.06 3.1 0.26 87 0.18 0 0.28 -2.35 83.02 100 0.02 5.84 10 8.59 85 1.06 3.1 0.26 87.2 0.1 0 0.33 -2.59 90.57 100 0.07 6.43 10 9.69 85 1.06 3.1 0.26 87.2 0.18 0 0.3 -2.59 83.02 100 0.04 6.51 10 1.01 85 1.06 3.1 0.26 81.9 1.01 1.79 0.27 3.65 4.72 42.26 0.01 3.2 10 2.59 85 1.06 3.1 0.26 82.9 1.01 0.86 0.26 2.47 4.72 72.26 0 3.69 10 2.96 85 1.06 3.1 0.26 85.5 0.94 0 0.29 -0.59 11.32 100 0.03 4.05 10 4.23 85 1.06 3.1 0.26 86.1 0.36 0 0.28 -1.29 66.04 100 0.02 4.84 10 5.33 85 1.06 3.1 0.26 83.7 0.22 0 0.26 1.53 79.25 100 0 5.14 10 7.25 85 1.06 3.1 0.26 85.6 0.18 0 0.31 -0.71 83.02 100 0.05 5.57 10 8.51 85 1.06 3.1 0.26 84.1 0.14 0 0.28 1.06 86.79 100 0.02 6.27 10 10.1 85 1.06 3.1 0.26 83.4 0.13 0 0.49 1.88 87.74 100 0.23 6.56

100

Stellenbosch University http://scholar.sun.ac.za

Appendix A (Experimental and Modelling results) Appendix A (Experimental and Modelling results) (Experimental and Modeling results)

Na addition to aqueous Cyanex 272 Ni-loading Nio aqueous Coo aqueous Feo aqueous Nao aqueous Ni aqueous Co aqueous Fe aqueous Na aqueous Ni extracted Co extracted Fe extracted phase after pH (V%) (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (%) (%) (%) extraction (g/l) 20 1.1 85 1.06 3.1 0.26 81.7 1 1.71 0.27 3.88 5.66 44.84 0.01 2.77 20 2.03 85 1.06 3.1 0.26 81.1 1 0.73 0.29 4.59 5.66 76.45 0.03 3.05 20 3.11 85 1.06 3.1 0.26 82.2 0.98 0.07 0.26 3.29 7.55 97.74 0 3.33 20 3.96 85 1.06 3.1 0.26 84.8 0.21 0 0.29 0.24 80.19 100 0.03 4.15 20 5.35 85 1.06 3.1 0.26 81.7 0.07 0 0.38 3.88 93.4 100 0.12 4.93 20 6.07 85 1.06 3.1 0.26 82.5 0.04 0 0.64 2.94 96.23 100 0.38 5.08 20 6.15 85 1.06 3.1 0.26 80.6 0.03 0 1.84 5.18 97.17 100 1.58 5.37 20 6.15 85 1.06 3.1 0.26 81.9 0.02 0 1.61 3.65 98.11 100 1.35 6.02 20 1.42 85 1.06 3.1 0.26 82.6 1 1.71 0.27 2.82 5.66 44.84 0.01 2.95 20 2.11 85 1.06 3.1 0.26 82.5 0.99 0.78 0.25 2.94 6.6 74.84 -0.01 3.06 20 3.24 85 1.06 3.1 0.26 86.8 0.9 0.06 0.25 -2.12 15.09 98.06 -0.01 3.5 20 4.38 85 1.06 3.1 0.26 84.7 0.21 0 0.28 0.35 80.19 100 0.02 4.24 20 5.88 85 1.06 3.1 0.26 85.9 0.06 0 0.29 -1.06 94.34 100 0.03 4.81 20 7.05 85 1.06 3.1 0.26 83.9 0.04 0 0.34 1.29 96.23 100 0.08 5.11 20 8.58 85 1.06 3.1 0.26 85.3 0.03 0 0.42 -0.35 97.17 100 0.16 5.48 20 9.63 85 1.06 3.1 0.26 87.7 0.03 0 0.46 -3.18 97.17 100 0.2 5.9 20 1.88 85 1.06 3.1 0.26 82.8 0.99 1.78 0.27 2.59 6.6 42.58 0.01 2.92 20 2.44 85 1.06 3.1 0.26 83.2 1.01 0.75 0.28 2.12 4.72 75.81 0.02 2.97 20 3.42 85 1.06 3.1 0.26 82.1 0.91 0.03 0.27 3.41 14.15 99.03 0.01 3.45 20 4.23 85 1.06 3.1 0.26 82.8 0.18 0 0.3 2.59 83.02 100 0.04 4.22 20 6.25 85 1.06 3.1 0.26 82 0.07 0 0.31 3.53 93.4 100 0.05 4.98 20 7.2 85 1.06 3.1 0.26 81.5 0.04 0 0.32 4.12 96.23 100 0.06 5.14 20 9.03 85 1.06 3.1 0.26 79.8 0.03 0 0.33 6.12 97.17 100 0.07 5.37 20 10.56 85 1.06 3.1 0.26 79.7 0.03 0 0.31 6.24 97.17 100 0.05 5.91

101

Stellenbosch University http://scholar.sun.ac.za

Appendix A (Experimental and Modelling results) Appendix A (Experimental and Modelling results) (Experimental and Modeling results)

Table A-6: Sodium addition to nickel product solution as a function of cobalt and iron extraction with pure Cyanex 272

Diluent: Kerosene Temperature (°C): 50 6.5.42 6.5.43 6.5.44 6.5.45 6.5.46 6.5.47 6.5.48 6.5.49 6.5.50 6.5.51 Agitation period (min): 5 6.5.52 6.5.53 6.5.54 6.5.55 6.5.56 6.5.57 6.5.58 6.5.59 6.5.60 6.5.61 O/A (ml/ml): 20/20 Na addition

Nio Coo Feo Nao Na to aqueous Cyanex 272 3 M NaOH Ni aqueous Co aqueous Fe aqueous Ni extracted Co extracted Fe extracted aqueous aqueous aqueous aqueous aqueous phase after pH (V%) (ml) (g/l) (g/l) (g/l) (%) (%) (%) (g/l) (g/l) (g/l) (g/l) (g/l) extraction (g/l) 10 0.15 82 1.02 3 0.56 79 1.01 2.03 0.59 2.94 -0.25 31.83 0.03 3.05 10 0.35 82 1.02 3 0.56 79 1.00 1.40 1.20 1.97 -0.25 52.52 0.64 3.34 10 0.5 82 1.02 3 0.56 78 0.99 0.83 1.68 2.50 0.02 71.64 1.12 3.37 10 0.65 82 1.02 3 0.56 77 0.96 0.21 2.20 3.05 2.34 92.77 1.64 3.40 10 0.75 82 1.02 3 0.56 78 0.83 0.04 2.58 1.31 15.16 98.62 2.02 3.81 10 0.85 82 1.02 3 0.56 76 0.62 0.02 2.92 3.38 36.32 99.31 2.36 4.24 10 1 82 1.02 3 0.56 75 0.29 0.02 3.34 3.96 70.00 99.30 2.78 4.90 10 1.3 82 1.02 3 0.56 73 0.16 0.05 4.26 5.19 83.21 98.23 3.70 5.21 10 1.75 82 1.02 3 0.56 71 0.08 0.03 5.68 5.84 91.43 98.91 5.12 6.20 10 2 82 1.02 3 0.56 66 0.07 0.02 6.09 11.46 92.41 99.27 5.53 6.82 10 2.25 82 1.02 3 0.56 66 0.08 0.02 6.64 10.46 91.23 99.26 6.08 6.82 10 2.75 82 1.02 3 0.56 63 0.06 0.01 8.19 12.61 93.28 99.62 7.63 6.82 20 0.15 82 1.02 3 0.56 79 0.95 2.04 0.55 2.94 5.70 31.49 0.00 3.07 20 0.35 82 1.02 3 0.56 78 0.94 1.23 1.23 3.21 5.77 58.28 0.67 3.09 20 0.5 82 1.02 3 0.56 76 0.94 0.69 1.88 5.00 5.07 76.43 1.32 3.10 20 0.65 82 1.02 3 0.56 78 0.93 0.25 2.41 1.79 5.40 91.40 1.85 3.31 20 0.75 82 1.02 3 0.56 75 0.82 0.03 2.52 5.11 16.18 98.96 1.96 3.52 20 0.85 82 1.02 3 0.56 75 0.54 0.01 2.80 4.65 44.54 99.65 2.24 3.74 20 1 82 1.02 3 0.56 74 0.13 0.02 3.56 5.24 86.55 99.30 3.00 4.53 20 1.3 82 1.02 3 0.56 73 0.05 0.00 4.40 5.19 94.75 100.00 3.84 4.84 20 1.75 82 1.02 3 0.56 70 0.03 0.00 5.78 7.16 96.79 100.00 5.22 5.37 20 2 82 1.02 3 0.56 69 0.03 0.00 6.60 7.44 96.75 100.00 6.04 5.76 20 2.25 82 1.02 3 0.56 68 0.03 0.00 6.97 7.74 96.71 100.00 6.41 5.84 20 2.75 82 1.02 3 0.56 63 0.02 0.01 7.82 12.61 97.76 99.62 7.26 6.46

102

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

Table A-7: Effect of various diluents on nickel pre-loading

Temperature (°C): 50 Agitation period (min): 5 6.5.62 6.5.63 6.5.64 6.5.65 O/A (ml/ml): 20/20

Cyanex NaOH/Nio Nio aqueous Ni aqueous Ni extracted Ni-loading Diluent pH 272 (V%) (mol/mol) (g/l) (g/l) (%) (g/l) Kerosene 10 0.21 10.30 9.65 5.17 0.53 5.7 Kerosene 10 0.43 10.30 8.91 11.36 1.17 6.14 Kerosene 10 0.85 10.30 6.49 33.86 3.49 6.6 Kerosene 10 1.71 10.30 2.36 74.80 7.71 7 Kerosene 20 0.21 10.30 9.33 8.31 0.86 5.4 Kerosene 20 0.43 10.30 8.17 18.72 1.93 5.64 Kerosene 20 0.85 10.30 6.31 35.70 3.68 6 Kerosene 20 1.71 10.30 1.96 79.07 8.15 6.9 Heptane 10 0.21 10.30 8.79 13.62 1.40 5.8 Heptane 10 0.43 10.30 8.6 14.45 1.49 5.9 Heptane 10 0.85 10.30 5.81 40.79 4.20 6.6 Heptane 10 1.71 10.30 1.64 82.49 8.50 7.1 Heptane 20 0.21 10.30 9.22 9.40 0.97 5.35 Heptane 20 0.43 10.30 8.03 20.12 2.07 5.6 Heptane 20 0.85 10.30 5.92 39.67 4.09 5.95 Heptane 20 1.71 10.30 2.07 77.90 8.03 6.6 Toluene 10 0.21 10.30 8.92 12.34 1.27 6 Toluene 10 0.43 10.30 7.95 20.91 2.15 6.4 Toluene 10 0.85 10.30 5.59 43.03 4.43 6.8 Toluene 10 1.71 10.30 1.72 81.64 8.41 7.3 Toluene 20 0.21 10.30 9.61 5.56 0.57 5.6 Toluene 20 0.43 10.30 7.23 28.07 1.87 5.95 Toluene 20 0.85 10.30 5.8 40.89 4.21 6.3 Toluene 20 1.71 10.30 1.97 78.97 8.14 6.9 Xylene 10 0.21 10.30 8.55 15.98 1.65 6.1 Xylene 10 0.43 10.30 7.92 21.21 2.19 6.45 Xylene 10 0.85 10.30 6.22 36.61 3.77 6.7 Xylene 10 1.71 10.30 1.64 82.49 8.50 7.2 Xylene 20 0.21 10.30 8.68 14.70 1.51 5.8 Xylene 20 0.43 10.30 8.42 16.24 1.67 5.9 Xylene 20 0.85 10.30 6.02 38.65 3.98 6.3 Xylene 20 1.71 10.30 1.77 81.10 8.36 6.9 Shellsol D70 10 0.21 10.35 9.17 10.29 1.07 5.7 Shellsol D70 10 0.43 10.35 8.23 18.50 1.91 6 Shellsol D70 10 0.85 10.35 6.16 37.51 3.88 6.6 Shellsol D70 10 1.70 10.35 1.87 80.13 8.29 7 Shellsol D70 20 0.21 10.35 9.44 7.65 0.79 5.31 Shellsol D70 20 0.43 10.35 8.3 17.80 1.84 5.58 Shellsol D70 20 0.85 10.35 6.44 34.67 3.59 6 Shellsol D70 20 1.70 10.35 1.89 79.91 8.27 6.8

103

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

Table A-8: Effect of various diluents on cobalt and iron extraction with Ni-Cyanex 272

Temperature (°C): 50 Agitation period (min): 5 6.5.66 6.5.67 6.5.68 6.5.69 6.5.70 6.5.71 O/A (ml/ml): 20/20 6.5.72 6.5.73

Cyanex Ni- Nio Coo Feo Ni Co Fe Ni Co Fe

Diluent 272 loading aqueous aqueous aqueous aqueous aqueous aqueous extracted extracted extracted (V%) (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (%) (%) (%) Kerosene 10 0.53 81.96 1.05 2.99 80.4 1.05 1.76 1.91 0.00 41.14 Kerosene 10 1.17 81.96 1.05 2.99 82.1 1.04 0.86 -0.16 0.95 71.24 Kerosene 10 3.49 81.96 1.05 2.99 81 0.48 0.01 1.18 54.29 99.67 Kerosene 10 7.71 81.96 1.05 2.99 79.9 0.16 0 2.52 84.76 100.00 Kerosene 20 0.86 81.96 1.05 2.99 83 1.02 1.69 -1.26 2.86 43.48 Kerosene 20 1.93 81.96 1.05 2.99 87.7 1 0.79 -6.99 4.76 73.58 Kerosene 20 3.68 81.96 1.05 2.99 81.9 0.22 0 0.08 79.05 100.00 Kerosene 20 8.15 81.96 1.05 2.99 83.2 0.02 0 -1.50 98.10 100.00 Heptane 0.27 1.40 81.96 1.05 2.99 79.8 1.03 1.76 2.64 1.90 41.14 Heptane 0.27 1.49 81.96 1.05 2.99 80.1 0.99 0.75 2.28 5.71 74.92 Heptane 0.27 4.20 81.96 1.05 2.99 82.1 0.32 0 -0.16 69.52 100.00 Heptane 0.27 8.50 81.96 1.05 2.99 82 0.11 0 -0.04 89.52 100.00 Heptane 0.54 0.97 81.96 1.05 2.99 81.5 1.02 1.59 0.57 2.86 46.82 Heptane 0.54 2.07 81.96 1.05 2.99 79.4 0.99 0.69 3.13 5.71 76.92 Heptane 0.54 4.09 81.96 1.05 2.99 85 0.23 0 -3.70 78.10 100.00 Heptane 0.54 8.03 81.96 1.05 2.99 84 0.02 0 -2.48 98.10 100.00 Toluene 0.27 1.27 81.96 1.05 2.99 80.1 0.99 1.72 2.28 5.71 42.47 Toluene 0.27 2.15 81.96 1.05 2.99 84 1.02 0.86 -2.48 2.86 71.24 Toluene 0.27 4.43 81.96 1.05 2.99 86.4 0.48 0 -5.41 54.29 100.00 Toluene 0.27 8.41 81.96 1.05 2.99 84.3 0.21 0 -2.85 80.00 100.00 Toluene 0.54 0.57 81.96 1.05 2.99 82.4 1.03 1.78 -0.53 1.90 40.47 Toluene 0.54 1.87 81.96 1.05 2.99 82.6 1 0.82 -0.77 4.76 72.58 Toluene 0.54 4.21 81.96 1.05 2.99 86.9 0.22 0 -6.02 79.05 100.00 Toluene 0.54 8.14 81.96 1.05 2.99 91.7 0.02 0 -11.87 98.10 100.00 Xylene 0.27 1.65 81.96 1.05 2.99 85.6 1 1.74 -4.43 4.76 41.81 Xylene 0.27 2.19 81.96 1.05 2.99 83.6 0.98 0.77 -1.99 6.67 74.25 Xylene 0.27 3.77 81.96 1.05 2.99 87.5 0.44 0 -6.75 58.10 100.00 Xylene 0.27 8.50 81.96 1.05 2.99 87.9 0.2 0 -7.24 80.95 100.00 Xylene 0.54 1.51 81.96 1.05 2.99 81.2 1 1.78 0.94 4.76 40.47 Xylene 0.54 1.67 81.96 1.05 2.99 84 1.03 0.81 -2.48 1.90 72.91 Xylene 0.54 3.98 81.96 1.05 2.99 84.7 0.25 0 -3.33 76.19 100.00 Xylene 0.54 8.36 81.96 1.05 2.99 86.6 0.02 0 -5.65 98.10 100.00 Shellsol D70 0.27 1.07 82.1 1.14 3.25 81.4 1.13 1.81 0.85 0.88 44.31 Shellsol D70 0.27 1.91 82.1 1.14 3.25 81.9 1.12 0.94 0.24 1.75 71.08 Shellsol D70 0.27 3.88 82.1 1.14 3.25 86.7 0.55 0 -5.60 51.75 100.00 Shellsol D70 0.27 8.29 82.1 1.14 3.25 82.7 0.29 0 -0.73 74.56 100.00 Shellsol D70 0.54 0.79 82.1 1.14 3.25 84.1 1.11 1.82 -2.44 2.63 44.00 Shellsol D70 0.54 1.84 82.1 1.14 3.25 82.1 1.11 0.79 0.00 2.63 75.69 Shellsol D70 0.54 3.59 82.1 1.14 3.25 82.5 0.38 0 -0.49 66.67 100.00 Shellsol D70 0.54 8.27 82.1 1.14 3.25 82.5 0.02 0 -0.49 98.25 100.00

104

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

Table A-9: High and low values of the pre-loading and extraction 24 factorial designs

Pre-loading O/A ratio Extractant concentration NaOH addition Temperature Run a ml organic ml aqueous b (V%) c 3 M ml d (°C) 1 -1 40 20 -1 10 -1 1 -1 30 2 1 10 20 -1 10 -1 1 -1 30 3 -1 40 20 1 20 -1 1 -1 30 4 1 10 20 1 20 -1 1 -1 30 5 -1 40 20 -1 10 1 2 -1 30 6 1 10 20 -1 10 1 2 -1 30 7 -1 40 20 1 20 1 2 -1 30 8 1 10 20 1 20 1 2 -1 30 9 -1 40 20 -1 10 -1 1 1 60 10 1 10 20 -1 10 -1 1 1 60 11 -1 40 20 1 20 -1 1 1 60 12 1 10 20 1 20 -1 1 1 60 13 -1 40 20 -1 10 1 2 1 60 14 1 10 20 -1 10 1 2 1 60 15 -1 40 20 1 20 1 2 1 60 16 1 10 20 1 20 1 2 1 60 Extraction O/A ratio Extractant concentration Ni loading Temperature Run a ml organic ml aqueous b (V%) c g/l d (°C) 1 -1 40 20 -1 10 -1 2.83 -1 30 2 1 10 20 -1 10 -1 2.83 -1 30 3 -1 40 20 1 20 -1 2.47 -1 30 4 1 10 20 1 20 -1 2.47 -1 30 5 -1 40 20 -1 10 1 4.76 -1 30 6 1 10 20 -1 10 1 4.76 -1 30 7 -1 40 20 1 20 1 4.66 -1 30 8 1 10 20 1 20 1 4.66 -1 30 9 -1 40 20 -1 10 -1 2.83 1 60 10 1 10 20 -1 10 -1 2.83 1 60 11 -1 40 20 1 20 -1 2.47 1 60 12 1 10 20 1 20 -1 2.47 1 60 13 -1 40 20 -1 10 1 4.76 1 60 14 1 10 20 -1 10 1 4.76 1 60 15 -1 40 20 1 20 1 4.66 1 60 16 1 10 20 1 20 1 4.66 1 60

105

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

Table A-10: Results for the pre-loading factorial design as shown in Table A-9

Diluent: Shellsol D70 Agitation period (min): 5 6.5.74 6.5.75 6.5.76

Nio Nao Ni Na Na addition to aqueous Ni-loading Run/Repeat aqueous aqueous aqueous aqueous phase after extraction pH (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) 1/1 9.76 0.02 5.66 3.17 1.91 3.15 6.11 1/2 9.76 0.02 5.69 3.17 1.89 3.15 6.1 1/3 9.76 0.02 5.33 3.24 2.08 3.22 6.14 2/1 9.76 0.02 4.92 2.71 9.19 2.69 6.82 2/2 9.76 0.02 5.34 2.91 8.31 2.89 6.83 2/3 9.76 0.02 5.14 3.17 8.73 3.15 6.81 3/1 9.76 0.02 5.38 3.23 2.05 3.21 5.63 3/2 9.76 0.02 5.33 3.14 2.08 3.12 5.63 3/3 9.76 0.02 5.35 3.07 2.07 3.05 5.62 4/1 9.76 0.02 5.63 3.05 7.70 3.03 6.46 4/2 9.76 0.02 5.31 3.18 8.38 3.16 6.46 4/3 9.76 0.02 5.35 2.74 8.28 2.72 6.46 5/1 9.76 0.02 1.46 5.26 4.08 5.24 6.94 5/2 9.76 0.02 1.42 5.60 4.10 5.58 6.94 5/3 9.76 0.02 1.41 5.53 4.10 5.51 6.95 6/1 9.76 0.02 1.07 5.66 17.16 5.64 7.84 6/2 9.76 0.02 0.65 5.86 18.09 5.84 7.83 6/3 9.76 0.02 0.63 5.97 18.13 5.95 7.87 7/1 9.76 0.02 1.34 5.66 4.14 5.64 6.33 7/2 9.76 0.02 1.39 5.80 4.11 5.78 6.3 7/3 9.76 0.02 1.45 5.76 4.08 5.74 6.29 8/1 9.76 0.02 1.12 5.73 17.05 5.71 7.28 8/2 9.76 0.02 1.11 5.95 17.09 5.93 7.3 8/3 9.76 0.02 1.13 5.85 17.03 5.83 7.29 9/1 9.76 0.02 5.31 2.96 2.09 2.94 6.25 9/2 9.76 0.02 5.43 3.29 2.03 3.27 6.26 9/3 9.76 0.02 5.42 3.14 2.03 3.12 6.23 10/1 9.76 0.02 5.27 3.01 8.44 2.99 6.68 10/2 9.76 0.02 5.44 3.12 8.10 3.10 6.71 10/3 9.76 0.02 4.80 2.81 9.43 2.79 6.65 11/1 9.76 0.02 5.38 2.76 2.05 2.74 5.65 11/2 9.76 0.02 5.40 3.15 2.04 3.13 5.66 11/3 9.76 0.02 5.40 3.17 2.05 3.15 5.63 12/1 9.76 0.02 5.40 3.21 8.19 3.19 6.56 12/2 9.76 0.02 5.32 3.24 8.35 3.22 6.53 12/3 9.76 0.02 5.39 3.27 8.21 3.25 6.58 13/1 9.76 0.02 1.46 5.63 4.08 5.61 6.82 13/2 9.76 0.02 1.39 5.77 4.12 5.75 6.85 13/3 9.76 0.02 1.38 5.91 4.12 5.89 6.86 14/1 9.76 0.02 0.93 5.01 17.48 4.99 7.35 14/2 9.76 0.02 0.66 5.32 18.07 5.30 7.45 14/3 9.76 0.02 0.90 4.98 17.55 4.96 7.25 15/1 9.76 0.02 1.22 5.52 4.21 5.50 6.42 15/2 9.76 0.02 1.41 5.31 4.11 5.29 6.41 15/3 9.76 0.02 1.19 5.92 4.22 5.90 6.43 16/1 9.76 0.02 0.91 5.36 17.51 5.34 7.16 16/2 9.76 0.02 0.86 5.80 17.62 5.78 7.18 16/3 9.76 0.02 0.99 5.30 17.34 5.28 7.12

106

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

Table A-11: Results for the extraction factorial design as shown in Table A-9

Diluent: Shellsol D70 Agitation period (min): 5 6.5.77 6.5.78 6.5.79 6.5.80 6.5.81 6.5.82

Nio Coo Feo Ni Co Fe Nickel Cobalt Iron

Run/Repeat aqueous aqueous aqueous aqueous aqueous aqueous extracted extracted extracted pH (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (%) (%) (%) 1/1 85.30 1.09 2.96 83.10 0.14 0.03 2.58 87.10 98.89 5.04 1/2 85.30 1.09 2.96 79.00 0.16 0.00 7.39 85.07 100.00 4.97 1/3 85.30 1.09 2.96 82.80 0.17 0.00 2.93 84.79 100.00 4.87 2/1 85.30 1.09 2.96 82.00 1.03 1.82 3.87 5.16 38.48 3.55 2/2 85.30 1.09 2.96 76.80 0.97 1.95 9.96 10.23 34.26 3.70 2/3 85.30 1.09 2.96 79.90 0.97 1.70 6.33 10.97 42.57 3.40 3/1 85.30 1.09 2.96 78.30 0.24 0.03 8.21 77.88 98.92 4.45 3/2 85.30 1.09 2.96 77.90 0.23 0.00 8.68 78.99 100.00 4.55 3/3 85.30 1.09 2.96 81.40 0.22 0.00 4.57 80.00 100.00 4.60 4/1 85.30 1.09 2.96 78.50 0.86 1.70 7.97 20.83 42.57 3.76 4/2 85.30 1.09 2.96 77.60 0.94 1.99 9.03 13.55 32.70 3.87 4/3 85.30 1.09 2.96 77.70 1.00 1.96 8.91 7.47 33.82 3.60 5/1 85.30 1.09 2.96 84.30 0.02 0.00 1.17 97.79 100.00 5.64 5/2 85.30 1.09 2.96 78.60 0.02 0.00 7.85 97.79 100.00 5.70 5/3 85.30 1.09 2.96 83.40 0.02 0.00 2.23 97.97 100.00 5.70 6/1 85.30 1.09 2.96 78.00 0.92 1.09 8.56 15.30 63.31 4.64 6/2 85.30 1.09 2.96 77.50 0.89 1.09 9.14 18.34 63.07 4.72 6/3 85.30 1.09 2.96 81.10 0.91 0.91 4.92 15.76 69.19 4.50 7/1 85.30 1.09 2.96 80.30 0.02 0.00 5.86 98.34 100.00 5.27 7/2 85.30 1.09 2.96 78.60 0.02 0.00 7.85 98.25 100.00 5.27 7/3 85.30 1.09 2.96 78.40 0.02 0.00 8.09 98.43 100.00 5.27 8/1 85.30 1.09 2.96 83.80 0.98 1.00 1.76 9.31 66.39 4.05 8/2 85.30 1.09 2.96 76.50 0.90 1.11 10.32 17.42 62.60 4.20 8/3 85.30 1.09 2.96 83.00 1.04 0.77 2.70 4.61 73.89 3.80 9/1 85.30 1.09 2.96 78.80 0.17 0.00 7.62 84.24 100.00 4.93 9/2 85.30 1.09 2.96 79.00 0.18 0.00 7.39 83.87 100.00 4.80 9/3 85.30 1.09 2.96 75.10 0.19 0.00 11.96 82.67 100.00 4.81 10/1 85.30 1.09 2.96 75.10 1.05 1.59 11.96 2.95 46.25 3.32 10/2 85.30 1.09 2.96 76.20 1.07 1.62 10.67 1.20 45.24 3.33 10/3 85.30 1.09 2.96 88.50 1.05 1.75 -3.75 3.23 40.81 3.27 11/1 85.30 1.09 2.96 71.20 0.19 0.00 16.53 82.40 100.00 4.24 11/2 85.30 1.09 2.96 72.50 0.22 0.00 15.01 80.09 100.00 4.32 11/3 85.30 1.09 2.96 72.00 0.20 0.00 15.59 82.03 100.00 4.27 12/1 85.30 1.09 2.96 79.60 1.08 1.45 6.68 0.46 51.01 3.04 12/2 85.30 1.09 2.96 79.60 1.05 1.35 6.68 3.23 54.49 3.06 12/3 85.30 1.09 2.96 77.20 1.09 1.54 9.50 0.00 48.14 3.10 13/1 85.30 1.09 2.96 78.90 0.02 0.00 7.50 98.16 100.00 5.68 13/2 85.30 1.09 2.96 80.80 0.01 0.00 5.28 99.08 100.00 5.71 13/3 85.30 1.09 2.96 78.70 0.01 0.00 7.74 98.89 100.00 5.66 14/1 85.30 1.09 2.96 88.20 1.06 0.73 -3.40 2.67 75.34 3.77 14/2 85.30 1.09 2.96 80.00 1.06 0.64 6.21 2.12 78.28 3.73 14/3 85.30 1.09 2.96 79.50 1.08 0.66 6.80 0.18 77.60 3.65 15/1 85.30 1.09 2.96 79.00 0.01 0.00 7.39 99.08 100.00 5.25 15/2 85.30 1.09 2.96 79.40 0.01 0.00 6.92 99.26 100.00 5.22 15/3 85.30 1.09 2.96 79.90 0.01 0.00 6.33 98.89 100.00 5.26 16/1 85.30 1.09 2.96 80.80 1.07 0.65 5.28 1.29 78.14 3.25 16/2 85.30 1.09 2.96 89.20 1.05 0.55 -4.57 3.59 81.28 3.27 16/3 85.30 1.09 2.96 90.00 1.00 0.55 -5.51 7.83 81.32 3.35

107

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

Table A-12: Effect of temperature on nickel loading of Cyanex 272

Cyanex 272 (V%): 10 Diluent: Kerosene 6.5.83 Agitation period (min): 5 6.5.84 O/A (ml/ml): 20/20

T (°C) 3 M NaOH (ml) Nio aqueous (g/l) Ni aqueous (g/l) Ni-loading (g/l) 25 1 5.8 1.5 4.23 25 1.36 5.8 0.9 4.84 25 1.64 5.8 0.1 5.69 25 1.5 8.6 3.1 5.27 25 1.82 9.7 2.2 7.30 25 2.27 9.7 1 8.59 50 0.25 6.15 5.41 0.66 50 0.5 6.15 4.3 1.51 50 0.75 6.15 3.48 2.4 50 1 6.15 2.15 3.19 50 1.3 6.15 1.25 4.23 50 0.25 10.5 9.84 0.54 50 0.5 10.5 8.48 1.81 50 0.75 10.5 7.68 2.53 50 1 10.5 6 4.20 50 1.3 10.5 5.08 5.09 50 1.6 10.5 3.36 6.87 50 2 10.5 1.74 8.59 50 2.25 10.5 0.73 9.69

108

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

Table A-13: Effect of Cyanex 272 concentration on nickel loading of Cyanex 272

T (°C): 50 Diluent: Kerosene 6.5.85 6.5.86 Agitation period (min): 5 6.5.87 6.5.88 O/A (ml/ml): 20/20

Nio aqueous Ni aqueous Cyanex 272 (V%) 3 M NaOH (ml) Ni-loading (g/l) pH (g/l) (g/l) 10 0.25 6.15 5.41 0.67 5.63 10 0.5 6.15 4.3 1.74 5.9 10 0.75 6.15 3.48 2.54 6.33 10 1 6.15 2.15 3.89 6.5 10 1.3 6.15 1.25 4.82 6.92 10 1.6 6.15 0.02 6.13 8.08 10 2 6.15 0 6.15 8.9 10 2.25 6.15 0.06 6.08 9 10 0.25 10.5 9.84 0.54 5.68 10 0.5 10.5 8.48 1.81 5.99 10 0.75 10.5 7.68 2.53 6.13 10 1 10.5 6 4.20 6.69 10 1.3 10.5 5.08 5.09 6.74 10 1.6 10.5 3.36 6.87 6.91 10 2 10.5 1.74 8.59 7.07 10 2.25 10.5 0.73 9.69 7.23 20 0.25 6.15 4.99 1.10 5.47 20 0.5 6.15 4.02 2.03 5.56 20 0.75 6.15 2.93 3.11 5.82 20 1 6.15 2.09 3.96 5.91 20 1.3 6.15 0.75 5.35 6.23 20 1.6 6.15 0.07 6.07 6.76 20 2 6.15 0 6.15 7.72 20 2.25 6.15 0 6.15 8.06 20 0.25 10.5 8.97 1.42 5.47 20 0.5 10.5 8.19 2.11 5.64 20 0.75 10.5 7 3.24 5.72 20 1 10.5 5.83 4.38 5.83 20 1.3 10.5 4.34 5.88 5.95 20 1.6 10.5 3.19 7.05 6.13 20 2 10.5 1.75 8.58 6.4 20 2.25 10.5 0.78 9.63 6.5

109

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

Table A-14: Effect of Cyanex 272 concentration on cobalt and iron extraction with Ni-Cyanex 272

T(°C): 50 Diluent: Kerosene 6.5.89 6.5.90 6.5.91 6.5.92 6.5.93 Agitation period (min): 5 6.5.94 6.5.95 6.5.96 6.5.97 6.5.98 O/A (ml/ml): 20/20 6.5.99 6.5.100 6.5.101 6.5.102 6.5.103 Cyanex Nio Coo Feo Ni Co Fe Nickel Cobalt Iron Ni-loading 272 aqueous aqueous aqueous aqueous aqueous aqueous extracted extracted extracted (g/l) (V%) (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (%) (%) (%) 10 0.67 85 1.06 3.1 81 1.03 1.8 4.71 2.83 41.94 10 1.74 85 1.06 3.1 79.9 1.02 0.85 6.00 3.77 72.58 10 2.54 85 1.06 3.1 83.8 1.03 0.04 1.41 2.83 98.71 10 3.89 85 1.06 3.1 78.4 0.37 0 7.76 65.09 100.00 10 4.82 85 1.06 3.1 80.2 0.22 0 5.65 79.25 100.00 10 6.13 85 1.06 3.1 77.5 0.21 0 8.82 80.19 100.00 10 6.15 85 1.06 3.1 74.5 0.22 0 12.35 79.25 100.00 10 0.54 85 1.06 3.1 78.2 1.01 1.81 8.00 4.72 41.61 10 1.81 85 1.06 3.1 81 1.01 0.87 4.71 4.72 71.94 10 2.53 85 1.06 3.1 84.4 0.93 0.05 0.71 12.26 98.39 10 4.20 85 1.06 3.1 87.4 0.36 0 -2.82 66.04 100.00 10 5.09 85 1.06 3.1 85.9 0.22 0 -1.06 79.25 100.00 10 6.87 85 1.06 3.1 87 0.18 0 -2.35 83.02 100.00 10 8.59 85 1.06 3.1 87.2 0.1 0 -2.59 90.57 100.00 10 9.69 85 1.06 3.1 87.2 0.18 0 -2.59 83.02 100.00 10 1.01 85 1.06 3.1 81.9 1.01 1.79 3.65 4.72 42.26 10 2.59 85 1.06 3.1 82.9 1.01 0.86 2.47 4.72 72.26 10 2.96 85 1.06 3.1 85.5 0.94 0 -0.59 11.32 100.00 10 4.23 85 1.06 3.1 86.1 0.36 0 -1.29 66.04 100.00 10 5.33 85 1.06 3.1 83.7 0.22 0 1.53 79.25 100.00 10 7.25 85 1.06 3.1 85.6 0.18 0 -0.71 83.02 100.00 10 8.51 85 1.06 3.1 84.1 0.14 0 1.06 86.79 100.00 10 10.10 85 1.06 3.1 83.4 0.13 0 1.88 87.74 100.00 20 1.10 85 1.06 3.1 81.7 1 1.71 3.88 5.66 44.84 20 2.03 85 1.06 3.1 81.1 1 0.73 4.59 5.66 76.45 20 3.11 85 1.06 3.1 82.2 0.98 0.07 3.29 7.55 97.74 20 3.96 85 1.06 3.1 84.8 0.21 0 0.24 80.19 100.00 20 5.35 85 1.06 3.1 81.7 0.07 0 3.88 93.40 100.00 20 6.07 85 1.06 3.1 82.5 0.04 0 2.94 96.23 100.00 20 6.15 85 1.06 3.1 80.6 0.03 0 5.18 97.17 100.00 20 6.15 85 1.06 3.1 81.9 0.02 0 3.65 98.11 100.00 20 1.42 85 1.06 3.1 82.6 1 1.71 2.82 5.66 44.84 20 2.11 85 1.06 3.1 82.5 0.99 0.78 2.94 6.60 74.84 20 3.24 85 1.06 3.1 86.8 0.9 0.06 -2.12 15.09 98.06 20 4.38 85 1.06 3.1 84.7 0.21 0 0.35 80.19 100.00 20 5.88 85 1.06 3.1 85.9 0.06 0 -1.06 94.34 100.00 20 7.05 85 1.06 3.1 83.9 0.04 0 1.29 96.23 100.00 20 8.58 85 1.06 3.1 85.3 0.03 0 -0.35 97.17 100.00 20 9.63 85 1.06 3.1 87.7 0.03 0 -3.18 97.17 100.00 20 1.88 85 1.06 3.1 82.8 0.99 1.78 2.59 6.60 42.58 20 2.44 85 1.06 3.1 83.2 1.01 0.75 2.12 4.72 75.81 20 3.42 85 1.06 3.1 82.1 0.91 0.03 3.41 14.15 99.03 20 4.23 85 1.06 3.1 82.8 0.18 0 2.59 83.02 100.00 20 6.25 85 1.06 3.1 82 0.07 0 3.53 93.40 100.00 20 7.20 85 1.06 3.1 81.5 0.04 0 4.12 96.23 100.00 20 9.03 85 1.06 3.1 79.8 0.03 0 6.12 97.17 100.00 20 10.56 85 1.06 3.1 79.7 0.03 0 6.24 97.17 100.00

110

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

Table A-15: Effect of temperature on the cobalt and iron extraction with Ni-Cyanex 272

Cyanex 272 (V%): 20 Diluent: Kerosene 6.5.104 6.5.105 6.5.106 6.5.107 6.5.108 Agitation period (min): 5 6.5.109 6.5.110 6.5.111 6.5.112 6.5.113 O/A (ml/ml): 20/20 6.5.114 6.5.115 6.5.116 6.5.117 6.5.118

Nio Coo Feo Ni Co Fe Nickel Cobalt Iron Ni-loading T(°C) aqueous aqueous aqueous aqueous aqueous aqueous extracted extracted extracted (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (%) (%) (%) 25 1.05 72 1 3.1 67 0.8 2.2 6.94 20.00 29.03 25 2.02 72 1 3.1 69 0.7 1.4 4.17 30.00 54.84 25 2.71 72 1 3.1 71 0.4 0.73 1.39 60.00 76.45 25 3.81 72 1.1 3.1 72 0.4 1.02 0.00 63.64 67.10 25 8.14 72 1.1 3.1 72 0 0.01 0.00 100.00 99.68 25 7.77 72 1.1 3.1 72 0 0.01 0.00 100.00 99.68 25 10.78 72 1.1 3.1 72 0 0.01 0.00 100.00 99.68 25 12.77 72 1.1 3.1 69 0 0.01 4.17 100.00 99.68 25 8.68 72 1.1 3.1 70 0 0.01 2.78 100.00 99.68 25 11.44 72 1.1 3.1 70 0 0.01 2.78 100.00 99.68 25 13.60 72 1.1 3.1 71 0 0.02 1.39 100.00 99.35 50 1.10 85 1.06 3.1 81.7 1 1.71 3.88 5.66 44.84 50 2.03 85 1.06 3.1 81.1 1 0.73 4.59 5.66 76.45 50 3.11 85 1.06 3.1 82.2 0.98 0.07 3.29 7.55 97.74 50 3.96 85 1.06 3.1 84.8 0.21 0 0.24 80.19 100.00 50 5.35 85 1.06 3.1 81.7 0.07 0 3.88 93.40 100.00 50 6.07 85 1.06 3.1 82.5 0.04 0 2.94 96.23 100.00 50 6.15 85 1.06 3.1 80.6 0.03 0 5.18 97.17 100.00 50 6.15 85 1.06 3.1 81.9 0.02 0 3.65 98.11 100.00 50 1.42 85 1.06 3.1 82.6 1 1.71 2.82 5.66 44.84 50 2.11 85 1.06 3.1 82.5 0.99 0.78 2.94 6.60 74.84 50 3.24 85 1.06 3.1 86.8 0.9 0.06 -2.12 15.09 98.06 50 4.38 85 1.06 3.1 84.7 0.21 0 0.35 80.19 100.00 50 5.88 85 1.06 3.1 85.9 0.06 0 -1.06 94.34 100.00 50 7.05 85 1.06 3.1 83.9 0.04 0 1.29 96.23 100.00 50 8.58 85 1.06 3.1 85.3 0.03 0 -0.35 97.17 100.00 50 9.63 85 1.06 3.1 87.7 0.03 0 -3.18 97.17 100.00 50 1.88 85 1.06 3.1 82.8 0.99 1.78 2.59 6.60 42.58 50 2.44 85 1.06 3.1 83.2 1.01 0.75 2.12 4.72 75.81 50 3.42 85 1.06 3.1 82.1 0.91 0.03 3.41 14.15 99.03 50 4.23 85 1.06 3.1 82.8 0.18 0 2.59 83.02 100.00 50 6.25 85 1.06 3.1 82 0.07 0 3.53 93.40 100.00 50 7.20 85 1.06 3.1 81.5 0.04 0 4.12 96.23 100.00 50 9.03 85 1.06 3.1 79.8 0.03 0 6.12 97.17 100.00 50 10.56 85 1.06 3.1 79.7 0.03 0 6.24 97.17 100.00

111

Stellenbosch University http://scholar.sun.ac.za

Appendix A (Experimental and Modelling results) Appendix A (Experimental and Modelling results) (Experimental and Modeling results)

Table A-16: Variation in O/A ratio and nickel loading for nickel, cobalt and iron extraction

Cyanex 272 (V%): 20 Diluent: Shellsol D70 6.5.119 6.5.120 6.5.121 6.5.122 6.5.123 6.5.124 6.5.125 6.5.126 6.5.127 6.5.128 6.5.129 Agitation period (min): 5 6.5.130 6.5.131 6.5.132 6.5.133 6.5.134 6.5.135 6.5.136 6.5.137 6.5.138 6.5.139 6.5.140 T(°C): 50 Na addition to Ni- Nio Coo Feo Nao Ni Co Fe Na Nickel Cobalt Iron Ni- Co- Fe- O/A aqueous phase loading aqueous aqueous aqueous aqueous aqueous aqueous aqueous aqueous extracted extracted extracted organic organic organic pH (ml/ml) after extraction (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (%) (%) (%) (g/l) (g/l) (g/l) (g/l) 5/20 0.56 90.1 1.12 2.98 0.08 86.6 1.07 2.64 0.06 3.88 4.38 11.58 14.56 0.2 1.38 -0.02 3.16 10/20 0.56 90.1 1.12 2.98 0.08 85.9 1.08 2.5 0.06 4.66 3.58 15.97 8.96 0.08 0.95 -0.02 3.44 15/20 0.56 90.1 1.12 2.98 0.08 90.3 1.09 2.14 0.06 -0.22 2.68 28.09 0.3 0.04 1.12 -0.02 3.18 20/20 0.56 90.1 1.12 2.98 0.08 91.1 1.05 1.95 0.06 -1.11 6.08 34.56 -0.44 0.07 1.03 -0.03 3.05 30/20 0.56 90.1 1.12 2.98 0.08 87 1 1.4 0.06 3.44 10.38 52.95 2.63 0.08 1.05 -0.02 3.1 40/20 0.56 90.1 1.12 2.98 0.08 82.2 1.03 1.13 0.05 8.77 8.23 62.18 4.51 0.05 0.93 -0.03 3.13 50/20 0.56 90.1 1.12 2.98 0.08 86.3 1.06 0.83 0.08 4.22 5.46 72.01 2.08 0.02 0.86 -0.01 3.14 60/20 0.56 90.1 1.12 2.98 0.08 87.9 1.06 0.53 0.07 2.44 5.19 82.08 1.3 0.02 0.82 -0.01 3.19 5/20 1.97 90.1 1.12 2.98 0.08 89.3 1.12 2.3 0.06 0.89 0 22.68 5.17 0 2.7 -0.03 3.11 10/20 1.97 90.1 1.12 2.98 0.08 87.1 1.09 1.83 0.1 3.33 2.86 38.49 7.97 0.06 2.29 0.02 3.08 15/20 1.97 90.1 1.12 2.98 0.08 86.1 1.08 1.27 0.06 4.44 3.76 57.32 7.3 0.06 2.28 -0.02 3.09 20/20 1.97 90.1 1.12 2.98 0.08 86 1.05 0.75 0.07 4.55 5.72 74.87 6.07 0.06 2.23 -0.01 3.16 30/20 1.97 90.1 1.12 2.98 0.08 90 1.04 0.04 0.09 0.11 7.25 98.79 2.03 0.05 1.96 0.01 3.52 40/20 1.97 90.1 1.12 2.98 0.08 90.1 0.26 0.01 0.08 0 76.39 99.53 1.97 0.43 1.48 0 4.34 50/20 1.97 90.1 1.12 2.98 0.08 92.4 0.06 0.01 0.07 -2.55 94.45 99.6 1.05 0.42 1.19 -0.01 4.76 60/20 1.97 90.1 1.12 2.98 0.08 95.2 0.05 0.01 0.1 -5.66 95.8 99.63 0.27 0.36 0.99 0.01 4.98 5/20 2.94 90.1 1.12 2.98 0.08 85.5 1.06 2.11 0.08 5.11 5.55 29.19 21.34 0.25 3.48 0 3.42 10/20 2.94 90.1 1.12 2.98 0.08 85 1.07 1.17 0.06 5.66 4.11 60.6 13.14 0.09 3.61 -0.02 3.16 15/20 2.94 90.1 1.12 2.98 0.08 85.2 1.07 0.6 0.05 5.44 4.38 79.83 9.47 0.07 3.17 -0.03 3.13 20/20 2.94 90.1 1.12 2.98 0.08 94.3 0.99 0.09 0.05 -4.66 11.18 96.88 -1.26 0.13 2.89 -0.03 3.67 30/20 2.94 90.1 1.12 2.98 0.08 89.5 0.1 0.05 0.1 0.67 91.14 98.32 3.34 0.68 1.95 0.02 4.6 40/20 2.94 90.1 1.12 2.98 0.08 89 0.05 0.04 0.07 1.22 95.71 98.83 3.49 0.54 1.47 -0.01 5.06 50/20 2.94 90.1 1.12 2.98 0.08 89.6 0.04 0.02 0.08 0.55 96.33 99.33 3.14 0.43 1.18 0 5.08 60/20 2.94 90.1 1.12 2.98 0.08 87.9 0.04 0.01 0.09 2.44 96.78 99.56 3.67 0.36 0.99 0.01 5.14

112

Stellenbosch University http://scholar.sun.ac.za

Appendix A (Experimental and Modelling results) Appendix A (Experimental and Modelling results) (Experimental and Modeling results)

Na addition to Ni- Ni Co Fe Na Ni Co Fe Na Nickel Cobalt Iron Ni- Co- Fe- O/A o o o o aqueous phase loading aqueous aqueous aqueous aqueous aqueous aqueous aqueous aqueous extracted extracted extracted organic organic organic pH (ml/ml) after extraction (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (%) (%) (%) (g/l) (g/l) (g/l) (g/l) 5/20 3.95 90.1 1.12 2.98 0.08 93.3 1.1 1.92 0.08 -3.55 1.52 35.67 -8.85 0.07 4.25 0 3.47 10/20 3.95 90.1 1.12 2.98 0.08 89.9 1.1 0.77 0.07 0.22 1.43 74.03 4.35 0.03 4.41 -0.02 3.21 15/20 3.95 90.1 1.12 2.98 0.08 90 1.02 0.08 0.1 0.11 9.03 97.45 4.08 0.13 3.87 0.02 3.62 20/20 3.95 90.1 1.12 2.98 0.08 94.7 0.21 0.03 0.08 -5.11 81.13 98.93 -0.65 0.91 2.95 -0.01 4.67 30/20 3.95 90.1 1.12 2.98 0.08 91.4 0.05 0.03 0.11 -1.44 95.17 99.16 3.08 0.71 1.97 0.03 5.06 40/20 3.95 90.1 1.12 2.98 0.08 91.3 0.04 0.02 0.09 -1.33 96.51 99.4 3.35 0.54 1.48 0.01 5.21 50/20 3.95 90.1 1.12 2.98 0.08 90.7 0.04 0.01 0.16 -0.67 96.69 99.66 3.71 0.43 1.19 0.08 5.27 60/20 3.95 90.1 1.12 2.98 0.08 92.3 0.04 0 0.12 -2.44 96.78 100 3.21 0.36 0.99 0.04 5.3 5/20 4.69 90.1 1.12 2.98 0.08 90.7 1.11 1.77 0.06 -0.67 0.63 40.6 2.29 0.03 4.84 -0.02 3.67 10/20 4.69 90.1 1.12 2.98 0.08 93.3 1.09 0.66 0.06 -3.55 2.86 78.02 -1.71 0.06 4.65 -0.02 3.83 15/20 4.69 90.1 1.12 2.98 0.08 87.2 0.55 0.05 0.07 3.22 50.54 98.26 8.56 0.75 3.9 -0.01 4.41 20/20 4.69 90.1 1.12 2.98 0.08 93.5 0.09 0.04 0.11 -3.77 91.68 98.56 1.29 1.03 2.94 0.03 4.89 30/20 4.69 90.1 1.12 2.98 0.08 93 0.04 0.01 0.1 -3.22 96.15 99.63 2.76 0.72 1.98 0.02 5.24 40/20 4.69 90.1 1.12 2.98 0.08 88 0.04 0 0.12 2.33 96.69 100 5.74 0.54 1.49 0.03 5.35 50/20 4.69 90.1 1.12 2.98 0.08 91.5 0.04 0 0.17 -1.55 96.78 100 4.13 0.43 1.19 0.09 5.41 60/20 4.69 90.1 1.12 2.98 0.08 89.6 0.03 0 0.15 0.55 96.96 100 4.86 0.36 0.99 0.07 5.46 5/20 5.73 90.1 1.12 2.98 0.08 88.8 1.09 1.47 0.13 1.44 2.24 50.57 10.93 0.1 6.03 0.05 3.77 10/20 5.73 90.1 1.12 2.98 0.08 93.1 1.06 0.13 0.08 -3.33 4.83 95.7 -0.27 0.11 5.7 0 3.6 15/20 5.73 90.1 1.12 2.98 0.08 88.1 0.17 0 0.07 2.22 85.06 100 8.39 1.27 3.97 -0.01 4.93 20/20 5.73 90.1 1.12 2.98 0.08 92.9 0.07 0 0.12 -3.11 94.01 100 2.93 1.05 2.98 0.04 5.1 30/20 5.73 90.1 1.12 2.98 0.08 89.3 0.04 0 0.12 0.89 96.33 100 6.26 0.72 1.99 0.04 5.33 40/20 5.73 90.1 1.12 2.98 0.08 90.8 0.04 0 0.15 -0.78 96.87 100 5.38 0.54 1.49 0.06 5.5 50/20 5.73 90.1 1.12 2.98 0.08 88 0.03 0 0.25 2.33 97.05 100 6.57 0.43 1.19 0.17 5.52 60/20 5.73 90.1 1.12 2.98 0.08 92.3 0.03 0 0.2 -2.44 97.14 100 4.99 0.36 0.99 0.12 5.58

113

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

Table A-17: Variation in O/A ratio and NaOH addition for nickel loading of Cyanex 272

Cyanex 272 (V%): 20 Diluent: Shellsol D70 6.5.141 6.5.142 6.5.143 Agitation period (min): 5 6.5.144 6.5.145 6.5.146 T (°C): 50

Nio Nao Na Ni - Ni- 3 M NaOH Ni aqueous O/A (ml/ml) aqueous aqueous aqueous extracted loading pH (ml) (g/l) (g/l) (g/l) (g/l) (%) (g/l) 80/3 2.5 88.4 0.4 7.4 34 84.65 2.81 5.99 80/3 3 88.4 0.4 2.7 34.9 93.89 3.11 6.4 80/3 4 88.4 0.4 0.6 33 98.42 3.26 6.95 80/3 5 88.4 0.4 0 26 100.00 3.32 7.28 80/3 5.5 88.4 0.4 0 26.2 100.00 3.32 7.37 80/3 6 88.4 0.4 0 26.4 100.00 3.32 7.48 80/4 2.5 88.4 0.4 16.9 27.8 68.93 3.05 5.86 80/4 3 88.4 0.4 11 33.9 78.22 3.46 6.05 80/4 4 88.4 0.4 1.1 30.6 97.51 4.31 6.6 80/4 5 88.4 0.4 0 27.7 100.00 4.42 7.01 80/4 5.5 88.4 0.4 0 28 100.00 4.42 7.24 80/4 6 88.4 0.4 0 25.9 100.00 4.42 7.4 80/5 2.5 88.4 0.4 23 21.7 60.97 3.37 5.78 80/5 3 88.4 0.4 17 25.7 69.23 3.83 5.82 80/5 4 88.4 0.4 7.3 30.7 85.14 4.70 6.28 80/5 5 88.4 0.4 1.5 32.5 96.61 5.34 6.67 80/5 5.5 88.4 0.4 0.4 30.8 99.05 5.47 6.96 80/5 6 88.4 0.4 0 31.2 100.00 5.53 7.2 80/6 2.5 88.4 0.4 31.5 20.9 49.52 3.28 5.76 80/6 3 88.4 0.4 25.3 24.2 57.07 3.78 5.86 80/6 4 88.4 0.4 13.9 28.4 73.79 4.89 6.17 80/6 5 88.4 0.4 6.3 29.9 86.93 5.76 6.42 80/6 5.5 88.4 0.4 2.8 31.1 93.93 6.23 6.65 80/6 6 88.4 0.4 0.9 32.6 97.96 6.50 6.9

Table A-18: Data for the stripping kinetics of nickel, cobalt and iron

Diluent: Shellsol D70 Cyanex 272 (V%) 20 Temperature (°C): 50 A/O (ml/ml): 20/20

Nio Coo Feo Ni Co Fe Ni Co Fe Agitation period H2SO4 organic organic organic aqueous aqueous aqueous stripped stripped stripped (s) (M) (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (%) (%) (%) 30 0.1 4.52 1.05 2.96 3.48 0.37 0.00 77.03 35.57 0.00 60 0.1 4.52 1.05 2.96 3.38 0.68 0.00 74.81 64.74 0.00 300 0.1 4.52 1.05 2.96 3.91 0.85 0.00 86.54 81.29 0.00 420 0.1 4.52 1.05 2.96 3.90 0.89 0.00 86.32 84.63 0.00 600 0.1 4.52 1.05 2.96 3.94 0.92 0.00 87.21 87.88 0.00 900 0.1 4.52 1.05 2.96 4.00 0.92 0.00 88.53 87.98 0.00 30 0.5 4.52 1.05 2.96 2.60 0.50 1.71 57.55 48.10 57.73 60 0.5 4.52 1.05 2.96 2.92 0.58 1.93 64.63 55.47 65.02 300 0.5 4.52 1.05 2.96 4.17 0.99 2.40 92.30 95.06 81.03 420 0.5 4.52 1.05 2.96 4.41 1.01 2.34 97.61 96.49 79.07 600 0.5 4.52 1.05 2.96 4.32 1.02 2.37 95.62 97.16 79.88 900 0.5 4.52 1.05 2.96 4.29 1.01 2.42 94.95 96.30 81.63

114

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

Table A-19: High and low values of the stripping 2 4 factorial design

O/A ratio Acid concentration Metal loading Temperature Run a ml organic ml aqueous b M c g/l Ni, Co, Fe d (°C) 1 -1 20 40 -1 0.1 -1 2.26, 0.52, 1.58 -1 30 2 1 20 10 -1 0.1 -1 2.26, 0.52, 1.58 -1 30 3 -1 20 40 1 0.5 -1 2.26, 0.52, 1.58 -1 30 4 1 20 10 1 0.5 -1 2.26, 0.52, 1.58 -1 30 5 -1 20 40 -1 0.1 1 4.67, 1.03, 3.14 -1 30 6 1 20 10 -1 0.1 1 4.67, 1.03, 3.14 -1 30 7 -1 20 40 1 0.5 1 4.67, 1.03, 3.14 -1 30 8 1 20 10 1 0.5 1 4.67, 1.03, 3.14 -1 30 9 -1 20 40 -1 0.1 -1 2.26, 0.52, 1.58 1 60 10 1 20 10 -1 0.1 -1 2.26, 0.52, 1.58 1 60 11 -1 20 40 1 0.5 -1 2.26, 0.52, 1.58 1 60 12 1 20 10 1 0.5 -1 2.26, 0.52, 1.58 1 60 13 -1 20 40 -1 0.1 1 4.67, 1.03, 3.14 1 60 14 1 20 10 -1 0.1 1 4.67, 1.03, 3.14 1 60 15 -1 20 40 1 0.5 1 4.67, 1.03, 3.14 1 60 16 1 20 10 1 0.5 1 4.67, 1.03, 3.14 1 60

115

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

Table A-20: Results for the stripping factorial design as shown in table 5-19

Diluent: Shellsol D70 Agitation period (min): 10 6.5.147 6.5.148 6.5.149 6.5.150 6.5.151 6.5.152

Nio Coo Feo Ni Co Fe Nickel Cobalt Iron Run/Repeat organic organic organic aqueous aqueous aqueous stripped stripped stripped pH (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (%) (%) (%) 11 2.26 0.52 1.58 1.08 0.25 0.05 95.72 96.82 6.20 1.98 12 2.26 0.52 1.58 1.03 0.24 0.06 91.29 91.86 6.96 1.80 13 2.26 0.52 1.58 1.14 0.23 0.06 101.04 87.67 7.97 1.77 21 2.26 0.52 1.58 4.21 1.00 0.00 93.29 95.58 0.00 2.67 22 2.26 0.52 1.58 4.12 0.95 0.00 91.29 90.82 0.00 2.68 23 2.26 0.52 1.58 4.16 0.99 0.00 92.18 94.53 0.00 2.64 31 2.26 0.52 1.58 1.10 0.26 0.70 97.50 98.73 88.80 1.47 32 2.26 0.52 1.58 1.07 0.25 0.70 94.84 94.91 89.06 1.42 33 2.26 0.52 1.58 1.08 0.24 0.66 95.72 91.86 83.62 1.40 41 2.26 0.52 1.58 4.20 0.97 1.73 93.07 92.34 54.84 1.51 42 2.26 0.52 1.58 4.01 0.91 1.62 88.86 87.10 51.27 1.47 43 2.26 0.52 1.58 3.96 0.93 1.53 87.75 88.81 48.23 1.48 51 4.67 1.03 3.14 1.83 0.37 0.07 78.36 71.11 4.59 1.88 52 4.67 1.03 3.14 1.95 0.39 0.06 83.50 75.37 3.76 1.87 53 4.67 1.03 3.14 1.84 0.38 0.03 78.79 73.04 2.17 1.89 61 4.67 1.03 3.14 4.96 0.01 0.00 53.10 0.44 0.00 5.59 62 4.67 1.03 3.14 4.93 0.01 0.00 52.78 0.44 0.00 5.97 63 4.67 1.03 3.14 5.42 0.01 0.00 58.02 0.39 0.00 5.67 71 4.67 1.03 3.14 2.08 0.44 1.27 89.07 85.44 81.02 1.50 72 4.67 1.03 3.14 2.08 0.45 1.33 89.07 87.96 84.84 1.42 73 4.67 1.03 3.14 2.00 0.42 1.24 85.64 81.76 79.17 1.40 81 4.67 1.03 3.14 8.47 1.86 1.50 90.68 89.90 23.90 1.54 82 4.67 1.03 3.14 8.37 1.76 1.47 89.61 85.25 23.46 1.51 83 4.67 1.03 3.14 8.65 1.78 1.40 92.60 86.12 22.29 1.47 91 2.26 0.52 1.58 1.13 0.27 0.00 100.16 101.77 0.00 1.81 92 2.26 0.52 1.58 1.30 0.26 0.00 107.25 98.34 0.00 1.74 93 2.26 0.52 1.58 1.23 0.24 0.00 109.02 93.01 0.00 1.77 101 2.26 0.52 1.58 4.29 0.95 0.00 95.06 90.63 0.00 2.87 102 2.26 0.52 1.58 4.10 0.96 0.00 90.85 91.67 0.00 2.85 103 2.26 0.52 1.58 4.36 0.99 0.00 96.61 94.63 0.00 2.88 111 2.26 0.52 1.58 1.19 0.26 0.59 105.47 97.20 75.14 1.41 112 2.26 0.52 1.58 1.11 0.26 0.55 98.38 100.63 68.94 1.34 113 2.26 0.52 1.58 1.14 0.25 0.52 101.04 96.82 65.78 1.30 121 2.26 0.52 1.58 4.45 1.02 2.09 98.60 97.58 66.07 1.50 122 2.26 0.52 1.58 4.36 0.95 1.90 96.61 90.24 60.12 1.40 123 2.26 0.52 1.58 4.30 1.01 2.09 95.28 96.34 66.07 1.33 131 4.67 1.03 3.14 2.23 0.47 0.00 95.49 90.87 0.00 2.18 132 4.67 1.03 3.14 2.13 0.44 0.00 91.21 85.83 0.00 2.09 133 4.67 1.03 3.14 2.16 0.45 0.00 92.50 87.57 0.00 2.00 141 4.67 1.03 3.14 5.69 0.01 0.00 60.91 0.24 0.00 5.04 142 4.67 1.03 3.14 5.31 0.00 0.00 56.85 0.10 0.00 5.18 143 4.67 1.03 3.14 5.60 0.00 0.00 59.95 0.19 0.00 5.20 151 4.67 1.03 3.14 2.31 0.48 1.26 98.92 93.77 80.32 1.57 152 4.67 1.03 3.14 2.39 0.49 1.27 102.34 94.35 80.57 1.48 153 4.67 1.03 3.14 2.27 0.49 1.23 97.21 94.94 78.09 1.49 161 4.67 1.03 3.14 9.41 2.00 2.73 100.74 96.78 43.39 1.62 162 4.67 1.03 3.14 9.20 1.94 2.65 98.49 94.16 42.20 1.57 163 4.67 1.03 3.14 9.02 1.93 2.80 96.56 93.29 44.54 1.58

116

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

Table A-21: Variation in O/A ratio and aid concentration for nickel, cobalt and iron stripping

Cyanex 272 (V%) 20 Diluent: Shellsol D70 6.5.153 6.5.154 6.5.155 6.5.156 6.5.157 6.5.158 Agitation period (min): 10 6.5.159 6.5.160 6.5.161 6.5.162 6.5.163 6.5.164 T(°C) 50

Nio Coo Feo Ni Co Fe Nickel Cobalt Iron A/O H2SO4 (M) organic organic organic aqueous aqueous aqueous stripped stripped stripped pH (ml/ml) (g/l) (g/l) (g/l) (g/l) (g/l) (g/l) (%) (%) (%) 10/20 0.025 2.96 0.75 2.08 1.19 0.01 0.00 20.02 0.47 0.00 5.75 15/20 0.025 2.96 0.75 2.08 1.01 0.00 0.00 25.55 0.30 0.00 5.88 20/20 0.025 2.96 0.75 2.08 1.02 0.00 0.00 34.44 0.40 0.00 5.79 30/20 0.025 2.96 0.75 2.08 1.12 0.00 0.00 56.88 0.00 0.00 5.68 45/20 0.025 2.96 0.75 2.08 1.04 0.00 0.00 78.71 0.30 0.00 5.44 60/20 0.025 2.96 0.75 2.08 0.92 0.14 0.00 93.50 55.75 0.00 4.08 10/20 0.05 2.96 0.75 2.08 3.29 0.00 0.00 55.51 0.20 0.00 5.41 15/20 0.05 2.96 0.75 2.08 3.17 0.01 0.00 80.21 0.50 0.00 5.36 20/20 0.05 2.96 0.75 2.08 2.74 0.40 0.00 92.49 54.02 0.00 3.8 30/20 0.05 2.96 0.75 2.08 1.76 0.45 0.00 88.94 89.85 0.00 2.42 45/20 0.05 2.96 0.75 2.08 1.24 0.30 0.00 94.21 89.35 0.00 2.16 60/20 0.05 2.96 0.75 2.08 0.99 0.23 0.00 100.39 90.25 0.00 2.04 10/20 0.075 2.96 0.75 2.08 3.94 0.00 0.00 66.52 0.20 0.00 5.54 15/20 0.075 2.96 0.75 2.08 3.50 0.60 0.00 88.72 60.47 0.00 3.71 20/20 0.075 2.96 0.75 2.08 2.74 0.69 0.00 92.39 92.52 0.00 2.43 30/20 0.075 2.96 0.75 2.08 1.78 0.44 0.00 90.16 88.04 0.00 2.1 45/20 0.075 2.96 0.75 2.08 1.25 0.27 0.00 94.97 82.43 0.00 1.98 60/20 0.075 2.96 0.75 2.08 1.00 0.23 0.00 101.30 92.66 0.00 1.91 10/20 0.1 2.87 0.75 2.10 5.36 0.02 0.00 93.25 1.20 0.00 4.77 15/20 0.1 2.87 0.75 2.10 3.71 0.95 0.00 96.78 94.58 0.00 2.5 20/20 0.1 2.87 0.75 2.10 2.80 0.71 0.00 97.65 94.98 0.00 2.24 30/20 0.1 2.87 0.75 2.10 1.69 0.42 0.00 88.34 83.01 0.00 2.1 45/20 0.1 2.87 0.75 2.10 1.25 0.28 0.00 98.19 82.91 0.00 1.91 60/20 0.1 2.87 0.75 2.10 0.99 0.24 0.00 103.44 95.38 0.00 1.82 10/20 0.25 2.87 0.75 2.10 5.30 1.29 0.01 92.29 86.07 0.14 2.06 15/20 0.25 2.87 0.75 2.10 3.70 0.91 0.78 96.64 90.59 27.74 1.75 20/20 0.25 2.87 0.75 2.10 2.88 0.72 0.87 100.27 95.65 41.23 1.69 30/20 0.25 2.87 0.75 2.10 1.89 0.48 0.69 98.73 95.58 49.48 1.7 45/20 0.25 2.87 0.75 2.10 1.38 0.31 0.45 101.87 93.39 48.15 1.6 60/20 0.25 2.87 0.75 2.10 1.05 0.23 0.32 104.48 93.39 45.76 1.58 10/20 0.5 2.87 0.75 2.10 6.00 1.48 2.95 104.48 98.64 70.26 1.6 15/20 0.5 2.87 0.75 2.10 3.76 0.95 2.21 98.11 94.28 78.97 1.46 20/20 0.5 2.87 0.75 2.10 2.75 0.70 1.71 95.88 93.25 81.46 1.45 30/20 0.5 2.87 0.75 2.10 1.98 0.49 1.20 103.17 98.57 85.87 1.4 45/20 0.5 2.87 0.75 2.10 1.24 0.32 0.74 97.17 95.48 79.58 1.39 60/20 0.5 2.87 0.75 2.10 1.02 0.25 0.57 106.57 98.97 81.51 1.38 10/20 0.75 2.89 0.72 2.03 5.70 1.33 3.05 98.64 93.01 75.05 1.5 15/20 0.75 2.89 0.72 2.03 3.88 0.95 2.43 100.71 99.37 89.68 1.43 20/20 0.75 2.89 0.72 2.03 2.87 0.72 1.82 99.13 100.35 89.23 1.42 30/20 0.75 2.89 0.72 2.03 2.02 0.45 1.23 104.89 94.76 90.71 1.39 45/20 0.75 2.89 0.72 2.03 1.28 0.33 0.78 99.80 103.14 85.73 1.36 60/20 0.75 2.89 0.72 2.03 0.96 0.26 0.57 99.65 108.59 83.78 1.34 10/20 1 2.89 0.72 2.03 5.79 1.39 3.00 100.11 97.41 73.77 1.4 15/20 1 2.89 0.72 2.03 3.70 0.90 2.82 95.94 94.55 104.09 1.39 20/20 1 2.89 0.72 2.03 2.99 0.74 1.94 103.56 103.14 95.23 1.37 30/20 1 2.89 0.72 2.03 1.97 0.50 1.26 102.03 104.61 93.14 1.3 45/20 1 2.89 0.72 2.03 1.34 0.34 0.90 103.93 105.34 99.23 1.29 60/20 1 2.89 0.72 2.03 0.96 0.27 0.61 99.96 104.82 90.12 1.27 10/20 2 2.89 0.72 2.03 5.76 1.30 3.51 99.65 91.05 86.28 <1.27 15/20 2 2.89 0.72 2.03 3.68 0.94 2.55 95.47 98.32 94.10 <1.28 20/20 2 2.89 0.72 2.03 2.86 0.73 1.94 99.02 101.75 95.18 <1.29 30/20 2 2.89 0.72 2.03 1.91 0.50 1.27 99.18 104.40 93.29 <1.30 45/20 2 2.89 0.72 2.03 1.33 0.34 0.85 103.31 106.60 93.47 <1.31 60/20 2 2.89 0.72 2.03 1.00 0.27 0.55 103.80 109.01 81.12 <1.32 10/20 4 2.89 0.72 2.03 5.85 1.44 4.03 101.24 100.35 99.04 <1.33 15/20 4 2.89 0.72 2.03 3.76 0.95 2.64 97.49 99.58 97.42 <1.34 20/20 4 2.89 0.72 2.03 3.00 0.76 1.94 103.80 105.80 95.28 <1.35 30/20 4 2.89 0.72 2.03 1.91 0.51 1.24 99.13 106.71 91.37 <1.36 45/20 4 2.89 0.72 2.03 1.34 0.36 0.80 104.40 106.92 88.94 <1.37 60/20 4 2.89 0.72 2.03 1.05 0.28 0.59 103.80 109.01 86.43 <1.38

117

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

A2 MODELLING AND SIMULATION

Table A-22: Constants solved my minimizing the square of the error in Microsoft Excel for the models developed in the process simulations sheet

K C b1 b2 b3 b4 b5 b6

Loading pH 14.42 1.000 -208.8 122.2 -17.07 -0.0003 - -

Nickel loading (g/l) 0.0106 ------

Extraction pH 7.432 1.335 0.2959 -0.5106 0.0205 -1.411 0.4319 -0.0348

Cobalt concentration in aqueous phase at equilibrium for the 0.0019 0.00002 - -3.018 - - - - extraction section (g/l)/Cobalt extraction (%)

Iron concentration in aqueous phase at equilibrium for the extraction 12.47 0.1246 -0.4788 -1.438 - - - - section (g/l)/Iron extraction (%)

Nickel concentration in aqueous phase at equilibrium for the stripping 12.84 0.1284 0.9778 -100.7 - - - - section (g/l)/Nickel stripped (%)

Cobalt concentration in aqueous phase at equilibrium for the stripping 0.0385 0.0004 6.827 -396.5 -0.3856 -0.0258 - - section (g/l)/Cobalt stripped (%)

Iron concentration in aqueous phase at equilibrium for the stripping 20.00 0.2005 3.527 -16.72 - - - - section (g/l)/Iron stripped (%)

8 8

6 6

4 Nickel loading (g/l) pH for nickel pH loadingfor 4 10 2 3 15 4 10 2 3 15 20 2 20 1 25 Na/Nio (mol/mol) O/A (ml/ml) Na/Ni (mol/mol) 1 25 O/A (ml/ml) 0 30 0 30

Figure A-1: Three dimensional graphs of the models developed for pH and nickel loading for the pre- loading section

118

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

1.5

5 1

3 0.5 Co in aqueous inCo aqueous pH for extraction pH for 2 4

8 extraction(g/l) after phase 0 6 2 4 0 0 2 1 2 4 2 6 Nickel loading (g/l) 0 0 O/A (ml/ml) Nickel loading (g/l) 3 O/A (ml/ml)

1 Fe in aqueous Fe in aqueous

0 phase after extraction (g/l) extraction after phase 0 0 2 1 4 2 Nickel loading (g/l) 6 3 O/A (ml/ml)

Figure A-2: Three dimensional graphs of the models developed for pH and cobalt and iron concentration in the aqueous phase at equilibrium for the extraction section

1.5

6 1 4

0.5

Co in aqueous inCo aqueous Ni in aqueous

0 0 phase after stripping (g/l) stripping after phase

phase after stripping (g/l) stripping after phase 0 0.4 1 0.25 0.2 0.2 A/O (ml/ml) 2 0.15 3 0.1 H2SO4(M) 2 2.5 0.05 1.5 3 0 H2SO4(M) 0 0.5 1 A/O (ml/ml)

Fe in aqueous Fe in aqueous 1

phase after stripping (g/l) stripping after phase 0 0 1 1 2 0.5 3 0 A/O (ml/ml) H2SO4(M)

Figure A-3: Three dimensional graphs of the models developed for nickel, cobalt and iron concentration at equilibrium in the aqueous phase for the stripping section

119

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

7.7 7.7

7.2 7.2

6.7 6.7 pH pH

6.2 6.2

5.7 5.7 0.8 1.3 1.8 2.3 2.8 3.3 3.8 4.3 0.8 1.3 1.8 2.3 2.8 3.3 3.8 4.3

NaOH/Nio (mol/mol) NaOH/Nio (mol/mol) 7.7 7.7

7.2 7.2

6.7 6.7 pH pH

6.2 6.2

5.7 5.7 0.8 1.3 1.8 2.3 2.8 3.3 3.8 4.3 0.8 1.3 1.8 2.3 2.8 3.3 3.8 4.3

NaOH/Nio (mol/mol) NaOH/Nio (mol/mol) 6.5 6.5 6.0 6.0 5.5 5.5 5.0 5.0 4.5 4.5 4.0 4.0 3.5 3.5

Nickel loading (g/l) loading Nickel 3.0 loading (g/l) Nickel 3.0 2.5 2.5 0.8 1.3 1.8 2.3 2.8 3.3 3.8 4.3 0.8 1.3 1.8 2.3 2.8 3.3 3.8 4.3 NaOH/Nio (mol/mol) NaOH/Nio (mol/mol)

6.5 6.5

6.0 6.0 5.5 5.5 5.0 5.0 4.5 4.5 4.0 4.0 3.5 3.5 Nickel loading loading (g/l) Nickel

Nickel loading loading (g/l) Nickel 3.0 3.0 2.5 2.5 0.8 1.3 1.8 2.3 2.8 3.3 3.8 4.3 0.8 1.3 1.8 2.3 2.8 3.3 3.8 4.3 NaOH/Nio (mol/mol) NaOH/Nio (mol/mol)

6.06.5 O/A = 26.67 (experimental) O/A = 26.67 (model) 5.05.5 4.5 O/A = 20 (experimental) O/A = 20 (model) 3.54.0

3.0(g/l)

Nickel Nickel 2.5

loading loading 2.0 O/A = 16 (experimental) O/A = 16 (model) 0.8O/A = 13.31.3 (experimental) NaOH/Nio1.8 2.3 (mol/mol) 2.8 O/A3.3 = 13.3 (model)3.8 4.3

Figure A-4: Model results at various O/A ratios as a function of NaOH addition for pH and nickel loading at equilibrium for the pre-loading section

120

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

6.0 6.0 6.0 5.5 5.5 5.5 5.0 5.0 5.0

4.5 4.5 pH

4.5 pH pH 4.0 4.0 4.0 3.5 3.5 3.5 3.0 3.0 3.0 0.25 1.25 2.25 3.25 0.25 1.25 2.25 3.25 0.25 1.25 2.25 3.25 O/A (ml/ml) O/A (ml/ml) O/A (ml/ml)

6.0 6.0 6.0 5.5 5.5 5.5 5.0 5.0 5.0

4.5 4.5

pH 4.5 pH pH 4.0 4.0 4.0 3.5 3.5 3.5 3.0 3.0 3.0 0.25 1.25 2.25 3.25 0.25 1.25 2.25 3.25 0.25 1.25 2.25 3.25 O/A (ml/ml) O/A (ml/ml) O/A (ml/ml)

1.2 1.2 1.2 1.0 1.0 1.0

0.8 0.8 0.8 0.6 0.6 0.6 Co (g/l) Co Co (g/l) Co 0.4 0.4 (g/l) Co 0.4 0.2 0.2 0.2 0.0 0.0 0.0 0.25 1.25 2.25 3.25 0.25 1.25 2.25 3.25 0.25 1.25 2.25 3.25 O/A (ml/ml) O/A (ml/ml) O/A (ml/ml) 1.2 1.2 1.2 1.0 1.0 1.0

0.8 0.8 0.8 0.6 0.6 0.6 Co (g/l) Co

0.4 (g/l) Co 0.4 (g/l) Co 0.4 0.2 0.2 0.2 0.0 0.0 0.0 0.25 1.25 2.25 3.25 0.25 1.25 2.25 3.25 0.25 1.25 2.25 3.25 O/A (ml/ml) O/A (ml/ml) O/A (ml/ml)

3.0 3.0 3.0

2.5 2.5 2.5

2.0 2.0 2.0

1.5 1.5 1.5 Fe (g/l) Fe Fe (g/l) Fe Fe (g/l) Fe 1.0 1.0 1.0

0.5 0.5 0.5

0.0 0.0 0.0 0.25 1.25 2.25 3.25 0.25 1.25 2.25 3.25 0.25 1.25 2.25 3.25 O/A (ml/ml) O/A (ml/ml) O/A (ml/ml)

121

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

3.0 3.0 3.0 2.5 2.5 2.5

2.0 2.0 2.0 1.5 1.5 1.5 Fe (g/l) Fe Fe (g/l) Fe

1.0 1.0 (g/l) Fe 1.0 0.5 0.5 0.5 0.0 0.0 0.0 0.25 1.25 2.25 3.25 0.25 1.25 2.25 3.25 0.25 1.25 2.25 3.25 O/A (ml/ml) O/A (ml/ml) O/A (ml/ml)

2.00 Ni loading = 0.56g/l (experimental) Ni loading = 0.56g/l (modell)

Ni loading = 1.97g/l (experimental) Ni loading = 1.97g/l (model) 1.00 Ni loading = 2.94g/l (experimental) Ni loading =2.94g/l (modell) Co (g/l) Co 0.00 Ni loading = 3.95g/l (experimental) Ni loading = 3.95g/l (model) 0.25 Ni 0.50loading =0.75 4.69g/l (experimental)1.00 1.25 1.50 1.75 2.00Ni loading 2.25 = 4.69g/l 2.50 (model) 2.75 3.00 Ni loading = 5.73g/l (experimental) O/A Ni loading = 5.73g/l (model)

Figure A-5: Model results at various nickel loadings as a function of O/A ratio for pH and cobalt and iron concentration in the aqueous phase at equilibrium for the extraction section

5.5 5.5 5.5

4.5 4.5 4.5

3.5 3.5 3.5

Ni (g/l) Ni 2.5 (g/l) Ni 2.5 2.5 Ni (g/l) Ni 1.5 1.5 1.5

0.5 0.5 0.5 0.00 0.05 0.10 0.15 0.20 0.25 0.00 0.05 0.10 0.15 0.20 0.25 0.00 0.05 0.10 0.15 0.20 0.25 H₂SO₄ (M) H₂SO₄ (M) H₂SO₄ (M)

5.5 5.5 5.5

4.5 4.5 4.5

3.5 3.5 3.5 Ni (g/l) Ni Ni (g/l) Ni 2.5 2.5 (g/l) Ni 2.5

1.5 1.5 1.5

0.5 0.5 0.5 0.00 0.05 0.10 0.15 0.20 0.25 0.00 0.05 0.10 0.15 0.20 0.25 0.00 0.05 0.10 0.15 0.20 0.25 H₂SO₄ (M) H₂SO₄ (M) H₂SO₄ (M)

1.4 1.4 1.4 1.2 1.2 1.2 1.0 1.0 1.0

0.8 0.8 0.8 0.6 0.6 0.6 Co (g/l) Co Co (g/l) Co 0.4 (g/l) Co 0.4 0.4 0.2 0.2 0.2 0.0 0.0 0.0 0.00 0.05 0.10 0.15 0.20 0.25 0.00 0.05 0.10 0.15 0.20 0.25 0.00 0.05 0.10 0.15 0.20 0.25

H₂SO₄ (M) H₂SO₄ (M) H₂SO₄ (M)

122

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

1.4 1.4 1.4 1.2 1.2 1.2 1.0 1.0 1.0

0.8 0.8 0.8 0.6 0.6 0.6

Co (g/l) Co 0.4 (g/l) Co 0.4 (g/l) Co 0.4 0.2 0.2 0.2 0.0 0.0 0.0 0.00 0.05 0.10 0.15 0.20 0.25 0.00 0.05 0.10 0.15 0.20 0.25 0.00 0.05 0.10 0.15 0.20 0.25

H₂SO₄ (M) H₂SO₄ (M) H₂SO₄ (M) 4.5 4.5 4.5 4.0 4.0 4.0 3.5 3.5 3.5 3.0 3.0 3.0 2.5 2.5 2.5 2.0 2.0 2.0

Fe (g/l) (g/l) Fe 1.5 (g/l) Fe 1.5 (g/l) Fe 1.5 1.0 1.0 1.0 0.5 0.5 0.5 0.0 0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

H₂SO₄ (M) H₂SO₄ (M) H₂SO₄ (M) 4.5 4.5 4.5 4.0 4.0 4.0 3.5 3.5 3.5 3.0 3.0 3.0 2.5 2.5 2.5 2.0 2.0 2.0

Fe (g/l) (g/l) Fe 1.5 (g/l) Fe 1.5 (g/l) Fe 1.5 1.0 1.0 1.0 0.5 0.5 0.5 0.0 0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

H₂SO₄ (M) H₂SO₄ (M) H₂SO₄ (M) A/O = 0.5 (experimental) A/O = 0.5 (model) 5

A/O = 0.75 (experimental) A/O = 0.75 (model)

pH A/O = 1 (experimental) A/O = 1 (model) A/O = 1.5 (experimental) A/O = 1.5 (model) 0 0.00 0.10A/O = 2.250.20 (experimental) 0.30 0.40 0.50 0.60 A/O0.70 = 2.250.80 (model) 0.90 1.00 A/O = 3 (experimental) H₂SO₄ (M) A/O = 3 (model)

Figure A-6: Model results at various A/O ratios as a function of H 2SO4 concentration for nickel, cobalt and iron concentration in the aqueous phase at equilibrium for the stripping section

123

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

A3 CONTINUOUS TESTS

Table A-23: Process conditions and results for the pre-loading (1st stage) and extraction (2 nd stage) test when generating a lower nickel loading

1st stage (pre-loading)

Nio - aqueous 6.5.165 g/l 10.74 6.5.166 6.5.167 6.5.168 6.5.169 6.5.170 Cyanex 272 6.5.171 V% 20 6.5.172 6.5.173 6.5.174 6.5.175 6.5.176 Diluent 6.5.177 - Shellsol6.5.178 D70 6.5.179 6.5.180 6.5.181 6.5.182 T 6.5.183 °C 40-50 6.5.184 6.5.185 6.5.186 6.5.187 6.5.188 O/A 6.5.189 ml/ml 1 6.5.190 6.5.191 6.5.192 6.5.193 6.5.194 Faq 6.5.195 ml/min 55 6.5.196 6.5.197 6.5.198 6.5.199 6.5.200 Forg 6.5.201 ml/min 70 6.5.202 6.5.203 6.5.204 6.5.205 6.5.206 FNaOH 6.5.207 ml/min 15 6.5.208 6.5.209 6.5.210 6.5.211 6.5.212 NaOH 6.5.213 g/l 40 6.5.214 6.5.215 6.5.216 6.5.217 6.5.218 Residence time in mixer 6.5.219 min 2.3 6.5.220 6.5.221 6.5.222 6.5.223 6.5.224 Residence time in settler 6.5.225 min 9.6 6.5.226 6.5.227 6.5.228 6.5.229 6.5.230 Ni - Na - Time after start-up aqueous aqueous pH Ni-loading (g/l)

(g/l) (g/l) 0 9.72 0.07 3.21 -1.95 5 9.69 0.07 5.97 -2.89 15 1.97 4.48 - 5.80 25 4.52 2.78 - 3.25 45 5.03 2.77 6.08 2.74 55 4.94 2.84 6 2.83 75 4.56 2.74 6.1 3.21 85 4.51 2.82 6.14 3.26 95 4.47 3.1 6.11 3.30 105 4.41 2.87 6 3.36 2nd stage (extraction)

Nio - aqueous 6.5.231 g/l 80 6.5.232 6.5.233 6.5.234 6.5.235 6.5.236 Coo - aqueous 6.5.237 g/l 1.05 6.5.238 6.5.239 6.5.240 6.5.241 6.5.242 Feo - aqueous 6.5.243 g/l 3.6 6.5.244 6.5.245 6.5.246 6.5.247 6.5.248 Nao - aqueous 6.5.249 g/l 0.08 6.5.250 6.5.251 6.5.252 6.5.253 6.5.254 Cyanex 272 6.5.255 6.5.256 V% 20 6.5.257 6.5.258 6.5.259 6.5.260 6.5.261 Diluent 6.5.262 6.5.263 - Shellsol6.5.264 D70 6.5.265 6.5.266 6.5.267 6.5.268 T 6.5.269 6.5.270 °C 40-50 6.5.271 6.5.272 6.5.273 6.5.274 6.5.275 O/A 6.5.276 6.5.277 ml/ml 1.5 6.5.278 6.5.279 6.5.280 6.5.281 6.5.282 Faq 6.5.283 6.5.284 ml/min 47 6.5.285 6.5.286 6.5.287 6.5.288 6.5.289 Forg 6.5.290 6.5.291 ml/min 70 6.5.292 6.5.293 6.5.294 6.5.295 6.5.296 Residence time in6.5.297 mixer 6.5.298 min 2.7 6.5.299 6.5.300 6.5.301 6.5.302 6.5.303 Residence time in6.5.304 settler 6.5.305 min 11.5 6.5.306 6.5.307 6.5.308 6.5.309 6.5.310 Na addition Ni - Co - Fe - Na – Ni Co Fe to aqueous Time after start-up aqueous aqueous aqueous aqueous pH extracted extracted extracted phase after (g/l) (g/l) (g/l) (g/l) (%) (%) (%) extraction (g/l) 0 - - - - 3.48 - - - - 5 6.6 0.09 0.22 0.14 3.15 91.75 91.44 93.89 0.06 15 44.5 0.245 0.652 0.12 - 44.38 76.69 81.88 0.04 25 69.1 0.153 0.375 0.09 - 13.63 85.44 89.58 0.01 45 81.7 0.103 0.242 0.08 4.17 -2.13 90.20 93.28 0 55 79.8 0.089 0.197 0.08 4.43 0.25 91.53 94.53 0 75 89.5 0.074 0.148 0.08 4.69 -11.88 92.96 95.89 0 85 90.4 0.063 0.099 0.07 4.97 -13.00 94.01 97.25 -0.01 95 83.2 0.058 0.043 0.07 5 -4.00 94.48 98.81 -0.01 105 83.5 0.051 0.058 0.08 5.05 -4.38 95.15 98.39 0

124

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

Table A-24: Process conditions and results for the pre-loading (1st stage) and extraction (2 nd stage) test when generating a higher nickel loading

1st stage (pre-loading)

Nio - aqueous 6.5.311 g/l 15 6.5.312 6.5.313 6.5.314 6.5.315 6.5.316 Cyanex 272 6.5.317 V% 20 6.5.318 6.5.319 6.5.320 6.5.321 6.5.322 Diluent 6.5.323 - Shellsol6.5.324 D70 6.5.325 6.5.326 6.5.327 6.5.328 T 6.5.329 °C 40-50 6.5.330 6.5.331 6.5.332 6.5.333 6.5.334 O/A 6.5.335 ml/ml 0.67 6.5.336 6.5.337 6.5.338 6.5.339 6.5.340 Faq 6.5.341 ml/min 77 6.5.342 6.5.343 6.5.344 6.5.345 6.5.346 Forg 6.5.347 ml/min 64 6.5.348 6.5.349 6.5.350 6.5.351 6.5.352 FNaOH 6.5.353 ml/min 18 6.5.354 6.5.355 6.5.356 6.5.357 6.5.358 NaOH 6.5.359 g/l 40 6.5.360 6.5.361 6.5.362 6.5.363 6.5.364 Residence time in mixer 6.5.365 min 2.01 6.5.366 6.5.367 6.5.368 6.5.369 6.5.370 Residence time in settler 6.5.371 min 8.5 6.5.372 6.5.373 6.5.374 6.5.375 6.5.376 Ni - Na - Time after start-up aqueous aqueous pH Ni-loading (g/l)

(g/l) (g/l) 0 15 0.105 3.35 -4.24 5 14.79 0.12 5.96 -3.93 15 8.805 2.82 6.43 5.05 25 8.715 2.865 6.43 5.19 45 8.85 2.925 6.4 4.98 55 8.85 2.82 6.5 4.98 75 8.85 2.955 6.54 4.98 85 8.73 2.925 6.5 5.16 2nd stage (extraction)

Nio - aqueous 6.5.377 g/l 82 6.5.378 6.5.379 6.5.380 6.5.381 6.5.382 Coo - aqueous 6.5.383 g/l 1.02 6.5.384 6.5.385 6.5.386 6.5.387 6.5.388 Feo - aqueous 6.5.389 g/l 3.07 6.5.390 6.5.391 6.5.392 6.5.393 6.5.394 Nao - aqueous 6.5.395 g/l 0.08 6.5.396 6.5.397 6.5.398 6.5.399 6.5.400 Cyanex 272 6.5.401 6.5.402 V% 20 6.5.403 6.5.404 6.5.405 6.5.406 6.5.407 Diluent 6.5.408 6.5.409 - Shellsol6.5.410 D70 6.5.411 6.5.412 6.5.413 6.5.414 T 6.5.415 6.5.416 °C 40-50 6.5.417 6.5.418 6.5.419 6.5.420 6.5.421 O/A 6.5.422 6.5.423 ml/ml 1.5 6.5.424 6.5.425 6.5.426 6.5.427 6.5.428 Faq 6.5.429 6.5.430 ml/min 43 6.5.431 6.5.432 6.5.433 6.5.434 6.5.435 Forg 6.5.436 6.5.437 ml/min 64 6.5.438 6.5.439 6.5.440 6.5.441 6.5.442 Residence time in6.5.443 mixer 6.5.444 min 2.99 6.5.445 6.5.446 6.5.447 6.5.448 6.5.449 Residence time in6.5.450 settler 6.5.451 min 12.6 6.5.452 6.5.453 6.5.454 6.5.455 6.5.456 Na addition to Ni - Co - Fe - Na – Ni Co Fe aqueous phase Time after start-up aqueous aqueous aqueous aqueous pH extracted extracted extracted after extraction (g/l) (g/l) (g/l) (g/l) (%) (%) (%) (g/l) 0 - - - - 3.28 - - - - 5 18.9 0.244 0.6 0.07 2.83 76.95 76.17 80.46 -0.01 15 61.7 0.241 0.55 0.11 3.2 24.76 76.46 82.08 0.03 25 72.8 0.111 0.24 0.12 3.94 11.22 89.16 92.18 0.04 45 80.5 0.053 0.16 0.13 4.6 1.83 94.82 94.79 0.05 55 84.5 0.043 0.05 0.12 4.91 -3.05 95.80 98.37 0.04 75 83.2 0.05 0.045 0.09 4.89 -1.46 95.12 98.53 0.01 85 85.2 0.044 0.05 0.12 4.9 -3.90 95.70 98.37 0.04

125

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

Table A-25: Process conditions and results when stripping the organic phase with the lower metal loading

1st stage (Ni and Co stripping) Ni - organic g/l 0.94 6.5.457 6.5.458 6.5.459 Co - organic g/l 0.37 6.5.460 6.5.461 6.5.462 Fe - organic g/l 1.37 6.5.463 6.5.464 6.5.465 Cyanex 272 V% 20 6.5.466 6.5.467 6.5.468 Diluent - Shellsol D70 6.5.469 6.5.470 6.5.471 T °C 40-50 6.5.472 6.5.473 6.5.474 A/O ml/ml 2.3 6.5.475 6.5.476 6.5.477 Faq ml/min 117 6.5.478 6.5.479 6.5.480 Forg ml/min 50 6.5.481 6.5.482 6.5.483 H2SO4 M 0.1 6.5.484 6.5.485 6.5.486 Residence time in mixer min 3.8 6.5.487 6.5.488 6.5.489 Residence time in settler min 16.2 6.5.490 6.5.491 6.5.492 Ni - aqueous Co - aqueous Fe – aqueous Ni stripped Co stripped Fe stripped Time after start-up (g/l) (g/l) (g/l) (%) (%) (%) 5 1.34 0.017 0.023 366.41 11.67 4.30 15 0.47 0.098 0.074 116.62 61.02 12.56 25 0.43 0.123 0.137 106.69 76.59 23.26 35 0.48 0.128 0.082 119.10 79.70 13.92 45 0.4 0.121 0.112 99.25 75.35 19.02 55 0.39 0.12 0.1 96.77 74.72 16.98 65 0.41 0.119 0.062 101.73 74.10 10.53 75 0.39 0.123 0.045 96.77 76.59 7.64 85 0.4 0.125 0.032 99.25 77.84 5.43 2nd stage (Fe stripping) Ni - organic g/l dependent on previous6.5.493 stage Co - organic g/l dependent on previous6.5.494 stage Fe - organic g/l dependent on previous6.5.495 stage Cyanex 272 V% 20 6.5.496 6.5.497 6.5.498 Diluent - Shellsol D70 6.5.499 6.5.500 6.5.501 T °C 40-50 6.5.502 6.5.503 6.5.504 A/O ml/ml 0.72 6.5.505 6.5.506 6.5.507 Faq ml/min 36 6.5.508 6.5.509 6.5.510 Forg ml/min 50 6.5.511 6.5.512 6.5.513 H2SO4 M 1 6.5.514 6.5.515 6.5.516 Residence time in mixer min 3.7 6.5.517 6.5.518 6.5.519 Residence time in settler min 15.7 6.5.520 6.5.521 6.5.522 Ni - aqueous Co - aqueous Fe - aqueous Ni stripped Co stripped Fe stripped Time after start-up (g/l) (g/l) (g/l) (%) (%) (%) 5 0 0 0 0.00 0.00 0.00 15 0 0 0 0.00 0.00 0.00 25 0 0 0 0.00 0.00 0.00 35 0.03 0.0113 0.22 2.13 2.01 10.67 45 0.03 0.023 0.404 2.31 4.44 21.29 55 0.04 0.031 0.426 3.08 5.99 22.45 65 0.05 0.036 0.503 3.85 6.96 26.50 75 0.06 0.045 0.535 4.62 8.70 28.19 85 0.07 0.045 0.64 5.39 8.70 33.72

126

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

Table A-26: Process conditions and results when stripping the organic phase with the higher metal loading

1st stage (Ni and Co stripping) Ni - organic g/l 3.07 6.5.523 6.5.524 6.5.525 Co - organic g/l 0.96 6.5.526 6.5.527 6.5.528 Fe - organic g/l 2.5 6.5.529 6.5.530 6.5.531 Cyanex 272 V% 20 6.5.532 6.5.533 6.5.534 Diluent - Shellsol D70 6.5.535 6.5.536 6.5.537 T °C 40-50 6.5.538 6.5.539 6.5.540 A/O ml/ml 2.3 6.5.541 6.5.542 6.5.543 Faq ml/min 117 6.5.544 6.5.545 6.5.546 Forg ml/min 50 6.5.547 6.5.548 6.5.549 H2SO4 M 0.1 6.5.550 6.5.551 6.5.552 Residence time in mixer min 3.8 6.5.553 6.5.554 6.5.555 Residence time in settler min 16.2 6.5.556 6.5.557 6.5.558 Ni - aqueous Co - aqueous Fe – aqueous Ni stripped Co stripped Fe stripped Time after start-up (g/l) (g/l) (g/l) (%) (%) (%) 5 0 0 0 0.00 0.00 0.00 15 0.5 0.229 0.199 38.00 55.49 18.55 25 0.62 0.272 0.262 47.12 65.91 24.42 35 0.59 0.253 0.222 44.84 61.30 20.70 45 0.69 0.287 0.193 52.44 69.54 17.99 55 1.14 0.238 0.22 86.64 57.67 20.51 65 1.18 0.247 0.262 89.69 59.85 24.42 75 1.17 0.229 0.216 88.93 55.49 20.14 85 1.16 0.234 0.214 88.17 56.70 19.95 95 1.19 0.24 0.229 90.45 58.15 21.35 105 1.2 0.242 0.208 91.21 58.64 19.39 115 1.19 0.247 0.25 90.45 59.85 23.31 2nd stage (Fe stripping) Ni - organic g/l dependent on previous6.5.559 stage Co - organic g/l dependent on previous6.5.560 stage Fe - organic g/l dependent on previous6.5.561 stage Cyanex 272 V% 20 6.5.562 6.5.563 6.5.564 Diluent - Shellsol D70 6.5.565 6.5.566 6.5.567 T °C 40-50 6.5.568 6.5.569 6.5.570 A/O ml/ml 1 6.5.571 6.5.572 6.5.573 Faq ml/min 50 6.5.574 6.5.575 6.5.576 Forg ml/min 50 6.5.577 6.5.578 6.5.579 H2SO4 M 1 6.5.580 6.5.581 6.5.582 Residence time in mixer min 6.4 6.5.583 6.5.584 6.5.585 Residence time in settler min 27 6.5.586 6.5.587 6.5.588 Ni - aqueous Co - aqueous Fe – aqueous Ni stripped Co stripped Fe stripped Time after start-up (g/l) (g/l) (g/l) (%) (%) (%) 5 0 0 0 0.00 0.00 0.00 15 0 0 0 0.00 0.00 0.00 25 0 0.01 0.066 0.00 1.04 2.64 35 0.18 0.106 0.592 5.86 11.01 23.65 45 0.44 0.196 1.059 14.33 20.35 42.31 55 0.61 0.251 1.358 19.87 26.06 54.25 65 0.51 0.257 1.506 16.61 26.69 60.17 75 0.51 0.275 1.593 16.61 28.56 63.64 85 0.58 0.295 1.724 18.89 30.63 68.88 95 0.59 0.31 1.728 19.22 32.19 69.04 105 0.59 0.295 1.737 19.22 30.63 69.40 115 0.5 0.32 1.724 16.29 33.23 68.88

127

Stellenbosch University http://scholar.sun.ac.za Appendix A (Experimental and Modelling results)

NaOH inventory chamber pH meters and flow meter Temperature meters

Pumps Stirrers

Mixers, settlers and aqueous Feed inventory over weirs containers and heating bath

Product batch

Figure A-7: Photograph of the mixer-settler setup

Stirrers

Aqueous over flow weirs

Mixers

Settlers

Figure A-8: Photograph of the agitators, mixers, settlers and aqueous overflow weirs

128

Stellenbosch University http://scholar.sun.ac.za Appendix B (Publications)

APPENDIX B (PUBLICATIONS)

Paper prepared for submission to refereed journal

Olivier, MC, Dorfling, C, and Eksteen, JJ, (2011), “Evaluating a solvent extraction process route incorporating nickel preloading of Cyanex 272 for the removal of cobalt and iron from nickel sulfate solutions”, Minerals Engineering

Refereed full length paper in the proceedings of international symposia

Olivier, MC, Dorfling, C, and Eksteen, JJ, (2011), “Extraction of cobalt(II) and iron(II) from nickel(II) solutions with nickel salts of Cyanex 272”, International Solvent Extraction Conference 2011, Santiago, Chile, 3-7 October 2011

Presentations at national conferences

Olivier, MC, Dorfling, C, and Eksteen, JJ, (2011), “Developing a solvent extraction process for the extraction of cobalt (II) and iron (II) from nickel (II) solutions with nickel salts of Cyanex 272”, Mineral Processing 2011, Western Cape Branch of SAIMM, Cape Town, South Africa, 4-5 August 2011

Olivier, MC, Dorfling, C, and Eksteen, JJ, (2010), “Developing a solvent extraction process for the separation of cobalt and iron from nickel sulfate solutions”, Mineral Processing 2010, Western Cape Branch of SAIMM, Cape Town, South Africa, 5-6 August 2010

129